48
$\begingroup$

This is intended to be a "humorous and friendly thread on how to ask a horrible question" (amoeba) which can help new users understand how to write good questions.

We hope you will contribute one "suggestion" per reply. Include (if you wish) its rank in the top ten reasons to close questions. But please add some explanation of why your suggestion is bad and (also if you wish) how to improve such questions.

One possible use of the replies in this thread will be targets of links provided in comments to closed questions.


Example

  1. The title is "Statistics question."

On a statistics site, this is meaningless. Good titles attract good readers. Make yours count. Use key words that clearly indicate what your question is about and how it might differ from similar ones.

(Often such titles reflect carelessness and lack of thought: they tend to be accompanied by questions elsewhere on our Top Ten list, such as copy-pasted homework. Such posts will usually be closed within minutes without further explanation.)


References

See our "TenFold" chat thread beginning at http://chat.stackexchange.com/transcript/message/30673850#30673850 for the first incarnation of this list. Thanks to Amoeba, Scortchi, user777, er, GeneralAbrial, um, Sycorax $\checkmark$, NickCox, Silverfish, Glen_b, usεr11852, Matthew, (and any others I may have overlooked) for your suggestions. Feel free to borrow from and expand on those ideas.

$\endgroup$
6
  • 8
    $\begingroup$ This could also be extended to "How to ask question that will never be answered?". $\endgroup$
    – Tim Mod
    Jul 2, 2016 at 17:32
  • $\begingroup$ @Tim Yeah, would be a good post. I'd never answer a question whose solution can be easily googled. Shows a significant lack of effort on the OP's part :) $\endgroup$
    – Dawny33
    Jul 5, 2016 at 4:48
  • 3
    $\begingroup$ I would love to see a variation on this question that is about "how to turn something that looks bad, but isnt, into something good" with emphasis on the transformation. If the "how to write a good question" was all powerful then .... ... perhaps effective reduction of bad questions could come from "how to transform". $\endgroup$ Oct 13, 2016 at 15:04
  • 1
    $\begingroup$ I have contributed to this thread, enjoyed doing so, and enjoyed reading other answers. I fear, however, that its main function is for experienced members to share horror stories of poor questions with each other, rather than as something that explains well and directly to naive or misguided questioners what they are doing wrong. $\endgroup$
    – Nick Cox
    Aug 25, 2022 at 11:18
  • $\begingroup$ @Nick The latter was the original intention, so I apologize if that wasn't clearly communicated. How would you suggest we fix things? $\endgroup$
    – whuber Mod
    Aug 25, 2022 at 13:28
  • 2
    $\begingroup$ I think the intention was utterly clear; no fault there. The problem lies in the tension between goals: it is all too easy to be wickedly humorous, but whether the result is friendly to people the other end is not so clear. As everyone contributing knows, it is not easy to distinguish between those who just want us to do their work for them and those out of their depth for some reason that deserves sympathy and support, whether statistically, scientifically, in terms of Stack Exchange or technical forum protocol, or even in elementary social skills. (I lack some of those too....) $\endgroup$
    – Nick Cox
    Aug 25, 2022 at 13:53

22 Answers 22

56
$\begingroup$

Questions which are just photographs of your homework sheet or old exam paper

enter image description here

Why these questions are closed

We have a policy on self-study questions, which says that:

It is okay to ask about homework. Homework is included in this self-study tag. This site exists to help people learn and provide a standard repository for questions in statistics and machine learning, both simple and complex, and this includes helping students.

However, we ask that you fulfil certain conditions, including:

  • Make a good faith attempt to solve the problem yourself first. If don't seem to be making a genuine attempt, your question might be voted down or closed.

  • Be honest about the source of the question. Do this by adding the self-study tag and mentioning whether it is for some class in the question text.

Just showing us a picture of the question doesn't show us what you've tried or where you're stuck, or indeed that you've made a good faith attempt at the question at all. Also, if you haven't typed anything, then we won't know what the original source of your question is. If your homework comes from a textbook, for instance, you should give a full reference (citation) to acknowledge its authorship.

Photographs of your question are also inconvenient for other reasons. They can be difficult to read, particularly if the quality of the photograph is poor. The text is inaccessible to users using screen readers. Moreover, the text is not searchable, either through our site search or external search engines — this goes against our objective of producing a high quality repository of statistics questions and answers that future readers can learn from. If your question can't be found, then other people will not be able to learn from it!

Sometimes people resort to taking a photograph of their homework because they do not feel able to copy the formatting. Our site has typesetting features for text and equations (using LaTeX) — see our editing help for further information. There are situations where photographs are appropriate, for instance if you want to ask what an unclear piece of notation means (and because it's not clear to you what it is, you can't typeset it), you are struggling with how to use LaTeX for an equation or table (some of our helpful users may edit your question to assist you with this), or you need to show us a graph or diagram. However, even in these cases it's best for your question not to just consist of a photograph! Type the question you have, and show us a picture of what you can't type.

$\endgroup$
9
  • 31
    $\begingroup$ As a legally blind guy, I really wish it were more commonly known that text should always be provided as text, not an image. $\endgroup$ Jul 2, 2016 at 13:48
  • 3
    $\begingroup$ There is also some information about using our mathjax / $\LaTeX$ markdown options here: Reviewing questions and applying LaTeX format. $\endgroup$ Jul 2, 2016 at 15:27
  • 16
    $\begingroup$ @Kodiologist Thanks -- it's very useful when visually impaired users identify themselves. I used to put at least briefly descriptive alt text on almost all images I put in my posts but I got the distinct impression nobody ever saw them -- so since it was taking time I could spend on other activities, I stopped doing it. You're the second user I'm aware of that it could possibly matter to, so I'll try to get back in the habit of doing it. When editing posts I sometimes try to convert images of text to text as well (but they're often poor self-study posts -- if it's likely to close, I won't). $\endgroup$
    – Glen_b
    Jul 3, 2016 at 1:08
  • 13
    $\begingroup$ That image is meta as meta gets... :) $\endgroup$ Jul 7, 2016 at 9:41
  • 3
    $\begingroup$ @J.M. It's even posted on the meta forum! $\endgroup$
    – Sycorax Mod
    Jan 19, 2017 at 1:21
  • 3
    $\begingroup$ I do love it when they at least type the question verbatim on the site, but add nothing more. The question is often phrased as a command (e.g. "find the value of $w$ that has the highest probability of generating $x$"). I often interpret it as a command by the OP, and I'm tempted to comment, "Yes sir!" $\endgroup$ Dec 5, 2017 at 18:31
  • 2
    $\begingroup$ Which flag should be used on such questions? (after OP refuses to make the changes himself) $\endgroup$
    – Jim
    May 16, 2018 at 8:02
  • 1
    $\begingroup$ @Jim That is an interesting point. There is an "off-topic" close reason for self-study questions that show insufficient effort which usually would cover such questions. $\endgroup$
    – Silverfish
    May 18, 2018 at 17:37
  • 1
    $\begingroup$ Fun fact: China apparently blocks imgur-hosted images, so users in China can't even see photographs. $\endgroup$ May 5, 2021 at 7:49
32
$\begingroup$

Questions which contain a large number of undefined abbreviations or acronyms

One of our close reasons is "Unclear what you are asking."

I'm comparing an XYZ model to a YYZ model using the ABC and DAR statistics. Using the LLV test on the OOP dataset, my results are not good. Perhaps I failed at TTR?

CrossValidated is a site for everyone, not just working or practicing statisticians or scientists. Questions that contain multiple (or sometimes even single) undefined abbreviations may be not understandable to a large number of otherwise interested readers.

As a rule, good scientific writing should avoid unnecessary abbreviations, if you only use a phrase once or twice, please take the time to type out the entire thing

I am confused by the formula for confidence interval given in my textbook. It seems to rely on the z-scores from the standard normal distribution, but I'm having trouble conceptualizing where the normal distribution comes into play in this situation (...describe situation...). I'd like to know, what is the relation between the z-score and the confidence interval in this situation?

You may think that confidence interval is such a common term that everyone will immediately recognize CI as a substitute, but consider a very beginner searching for an answer to their question:

Search: Confidence intervals and z scores

If you have abbreviated to CI all your references to the subject of your question, this user will never uncover your question or answer.

If you must repeatedly refer to a complex multi word term in your question or answer, please define your abbreviation or acronym on first use

I'm wondering about how Hyper-Complex Kahler Manifolds with a Holomorphic One Form (CKMwaH1F) ...

One last point that follows from the previous. Please avoid abbreviations completely in question titles. Titles are meant to be consumed quickly and give a summary of content to tell the reader if they have possible interest in the question, you are only doing yourself a disservice if you alienate a few readers by using unknown abbreviations.

Of course, there are counterexamples to this advice. Some abbreviations have become so standardized in our common language that the abbreviation/acronym has become the name, ANOVA being a prime example. This poster urges you to use your best discretion, but lean towards utmost sympathy to your readers. When in doubt, write it out.

$\endgroup$
4
  • 5
    $\begingroup$ Re the final paragraph - some abbreviations are okay (ANOVA for instance is rarely written unabbreviated!) so perhaps the best guidance is slightly more subtle than a blanket preference for everything to be written in full $\endgroup$
    – Silverfish
    Jul 2, 2016 at 19:17
  • 5
    $\begingroup$ (Of course, if you're someone who doesn't know much about statistics, or you've only seen an idea once, you may have no idea if the "usual" way it is written is abbreviated or not! So I can sympathise with posters who struggle with such conventions. But since so many TLAs are ambiguous, encouraging to write acronyms and abbreviations in full when in doubt, at least at their first mention, is sensible advice.) $\endgroup$
    – Silverfish
    Jul 2, 2016 at 22:34
  • $\begingroup$ @Silverfish I took a shot at making your point in my answer. Feel free to edit to better capture your intent! $\endgroup$ Jul 2, 2016 at 23:10
  • $\begingroup$ That looks good to me! But if I can come up with a more elegant way to rephrase it, I'll come back to it. $\endgroup$
    – Silverfish
    Jul 2, 2016 at 23:12
28
$\begingroup$

Not carefully developing a question to ask of your data

As Scortchi put it (his ellipsis):

[…] How can I analyze these data?

This quip exemplifies questions that describe a dataset and then ask what data-analytical, statistical, mathematical, programmatical, or machine-learning-ical tools to throw at the data—"What do I do with the data?"—without being reasonably specific about what the asker wants to achieve with data analysis. Data analysis is not a machine where you stick data in one end and get an APA-formatted results section out the other. It is a set of tools to help you answer questions about data. You need to decide on some questions to ask before the tools can do you any good.

To avoid this problem, try to do two things whenever you ask a question:

  1. Provide broader context. Why are you looking at this data to begin with? What were you hoping to learn, at the big-picture level?

  2. Focus your question enough to ask about a particular aspect of the data. Which variable (or variables) do you think of as the outcome? Are you interested in building a predictive model, identifying causes, summarizing a large dataset, searching for certain cases, or something else?

$\endgroup$
1
  • 1
    $\begingroup$ OMG YES. $\endgroup$
    – Alexis
    Sep 20, 2017 at 19:30
27
$\begingroup$

I personally like the "questions":

I have a dataset consisting of vectors of numbers and I want to predict values from other vector of numbers. Any help will be appreciated. Thx

Why it gets closed?

First of all, what is the question in here? Is there any? Nothing was asked, so it's hard to answer anything.

$\endgroup$
2
  • 1
    $\begingroup$ My problem when folks ask questions like this is that they come across as an amazingly passive/half-dead request for help. In terms of effective response (think Gladwell/Outliers) the most effective response can be elicited from the most complete, direct, and non-passive approach. $\endgroup$ Oct 13, 2016 at 15:01
  • $\begingroup$ @EngrStudent I don't disagree, but cultures vary in how they see this - and in some, being outspoken can be perceived as aggressive or obnoxious. Definitely agree that these need to be closed, though. $\endgroup$
    – mkt
    Aug 13, 2018 at 14:04
26
$\begingroup$

"I am working on my dissertation..."

Why is this a negative point? It may seem entirely natural as a lead-in, but it's a bad start.

We think statistically (what else did you expect?). A large fraction of questions starting like that turn out to be poor questions. Your question isn't doomed by such a start, but you should want to create a good first impression. You won't want to lose readers because they have previously found such questions unappealing and back off immediately.

Why is it bad?

  • We don't need to know that. It is not important information. The important information is the statistical question you want to ask. Whether that question arises in coursework, in preparing a dissertation or thesis, or in working towards a report or research paper is not crucial at all. Such context is a distraction on a par with whether it's Tuesday or the weather is good where you are. Tell us about your question: what you understand and what you need to know.

  • Whether you are a student doesn't make a difference. Often it creates the expectation that you are going to plead -- or at least to imply -- that you're a student in special need of support. But the forum doesn't care who you are. There isn't a queue you can jump by being needier than anyone else. We have no way of judging who needs an answer most. We just care about whether you have a good question.

These may seem small points, even petty, but your question is jostling for attention with many others, so you should want it to look good.

There are ways to make this kind of question even less attractive, so we will pile on with those too.

  • Don't imply that you don't respect statistics. Never give the impression that you regard the statistical part of what you're doing as boring and unnecessary technical stuff, a hurdle you're obliged to jump before you can finish your degree. You may think that quietly to yourself, but special pleading that you're not really that statistically-minded won't make your question more attractive or easier to answer. You can and should be honest about your technical level, but should feel no need to apologize for it. Perhaps you never took the right courses or they were not taught in a way that helped you. Just don't convey a lack of respect for statistics. If you don't respect what we do, don't expect copious and willing help.

  • Don't try emotional blackmail: it backfires. Never plead urgency, desperation, or your earnest need to finish. Don't claim a deadline tomorrow or that you've spent hours watching videos and Googling without any result except getting more frustrated and more confused. All this may be completely true, but it doesn't make a question a better question if you add layers of pleading or attempts at tweaking our heartstrings. We've been there ourselves. We know that there are bad reasons too for being in a mess, which we should not need to spell out.

It should seem obvious: Just ask the statistical question clearly, concisely, and concretely. No need for personal history, special pleading, and least of all for thinly veiled emotional blackmail.

$\endgroup$
5
  • 7
    $\begingroup$ +1. I totally agree in the emotional blackmail part. I find it totally irrational. $\endgroup$
    – usεr11852
    Jan 3, 2017 at 23:53
  • 5
    $\begingroup$ I think there are times when "I am working on my dissertation" can be relevant (such as to clarify what looks as if it might be simple homework is actually something else; I've seen this with some dissertation posts where the author tries to turn their problem into some artificial situation - often obscuring necessary information out of ignorance - it often ends up looking like a mangled homework question). There are also times when it might affect the kinds of advice offered (if it's a dissertation there will typically be different expectations on the asker than if it's for a job. ... ctd $\endgroup$
    – Glen_b
    Apr 12, 2017 at 2:15
  • $\begingroup$ ctd ... I agree you shouldn't ever need to lead with it. I completely agree with the rest of it and the general advice (especially about asking the statistical question clearly, concisely, and concretely). $\endgroup$
    – Glen_b
    Apr 12, 2017 at 2:17
  • 2
    $\begingroup$ @Glen_b I agree with you but the crucial detail is whether the answer would be better revised. Across countries and disciplines a dissertation can be for anything from first degree or diploma to doctorate. Expectations vary from using some elementary procedures to showing familiarity with frontier-research technique. I have seen many posts with the added flavour of "completely new to statistics" and "using X" where X can be zero-inflated negative binomial regression, mixed models, etc., etc. The central point for me is that whatever the personal context, what is good advice shouldn't vary. $\endgroup$
    – Nick Cox
    Apr 12, 2017 at 6:25
  • $\begingroup$ I should make it clear that I don't think the answer requires revision; I previously upvoted it as it stood. I just sought to note the possibility of occasional relevance of it being a thesis. $\endgroup$
    – Glen_b
    May 31, 2017 at 23:50
26
$\begingroup$

What was the chance of this strange personal coincidence?

I am perplexed; I just ended a relationship with a guy and 1 month later I meet another guy who has the same first name and the same first initial in the last name. Example : Torrence Tolley (1st guy) & Torrence Tiffen (2nd guy). What is the universe trying to tell me?

Yes, that's a real post from our site!--but there have been plenty of others like it.

Strange and unusual things happen all the time:

Million-to-one chances crop up nine out of ten times.

--Terry Pratchett

For an understandable explanation of the "law of very large numbers" expressed by Terry Pratchett, read Gina Kolata's NY Times article about research by Persi Diaconis and Fred Mosteller, "1-in-a-Trillion Coincidence, You Say? Not Really, Experts Find". Former Royal Statistical Society president David Hand has written an entire book on this "Improbability Principle" aimed at the layperson.

$\endgroup$
5
  • 4
    $\begingroup$ If we keep getting these then it might be worthwhile to duplicate, in some form, the A Very Lucky Wind story by RadioLab as a canonical answer and dupe all of the others. That way we don't need to answer all of these questions and the asker can get something useful. $\endgroup$
    – Erik
    Nov 30, 2016 at 21:05
  • $\begingroup$ True story: I met a man in Chile , he told this true story: many years ago, in his youth, he was in la Serena and had booked a flight to Santiago. He had some important business in Santiago, so was angry when it showed at the airport there was overbooking. Everybody busy, he lost. Angry looking at the plane taking off, disappearing at the horizon --- just to fall down in flames, no survivors. Not so angry then ... I use this story when teaching probability ... $\endgroup$ Nov 27, 2017 at 17:37
  • 1
    $\begingroup$ @kjetilbhalvorsen According to Wikipedia, that plane crashed while landing in la Serena, not after the departure. $\endgroup$
    – amoeba
    Sep 19, 2018 at 13:35
  • $\begingroup$ @amoeba: Interesting, but I told the story as I remember I was told it ... I could have messed up details in my head. Would not be the first time. $\endgroup$ Sep 19, 2018 at 13:39
  • 1
    $\begingroup$ Well, that was actually a very good question indeed! $\endgroup$
    – David
    Jul 5, 2019 at 10:41
24
$\begingroup$

QUESTIONS THAT ARE TYPED IN ALL CAPS

Why these questions are closed

Western languages differentiate between uppercase and lowercase letters. There are commonly agreed rules on when to use which. These rules are well ingrained in almost all speakers of English. The result is that, yes, we can read text that is in ALL CAPS, but it requires additional effort, simply because it does not adhere to commonly agreed rules.

If you want someone to help you, it is a good idea to make helping you as painless as possible. This includes (but is not limited to) making the minuscule extra effort of formatting your question more or less correctly.

Note that we understand that many CV users are not native speakers of English. We do not expect letter-perfect command of English. However, not typing your question in ALL CAPS should be possible for anyone who is using the internet.

One additional point: in the early days of the internet, typing in ALL CAPS was the visual equivalent of screaming. Many users here are old enough to remember that time and have the same reaction to a question in ALL CAPS that they would have to a question that was literally screamed at them.

Figure

Visit https://www.comicskingdom.com/shared_comics/458fbe98-97e3-426c-86c4-404c2c150493

$\endgroup$
9
  • 8
    $\begingroup$ (+1). I would still add that, while not a reason for immediate closing as English is indeed not most users' native language, an excessive number typos reflects that little effort has gone into the question, which increases the chances that nobody will make the effort to provide an answer either (maybe we need a sister thread "Top Ten List of Reasons not to Get an Anwswer on Your Question"?). $\endgroup$ Jul 4, 2016 at 8:17
  • 3
    $\begingroup$ @ChristophHanck +1 for planting a typo in the the suggested thread title. $\endgroup$
    – amoeba
    Jul 4, 2016 at 9:09
  • 1
    $\begingroup$ Oh, that is embarrassing in this context :-). $\endgroup$ Jul 4, 2016 at 9:13
  • 2
    $\begingroup$ That's OK @ChristophHanck, in another answer someone has an unexplained abbreviation. We are all human. $\endgroup$
    – mdewey
    Jul 4, 2016 at 10:51
  • 2
    $\begingroup$ Well, it gets worse than that - as it happens, my comment on the sister thread is a duplicate of @Tim's under the original post, while my answer suggested there should be fewer...duplicates on CV. $\endgroup$ Jul 5, 2016 at 11:46
  • 1
    $\begingroup$ "in the early days of the internet, typing in ALL CAPS was the visual equivalent of screaming" - interesting, I see it used in many forums in this way to this day and I interpret it this way - and I am not old enough to remember early days of the internet. Though maybe some interpret "before FB" as "early days of the internet", for me it would be "before Internet search was useful". $\endgroup$ Jul 6, 2016 at 22:05
  • 10
    $\begingroup$ "In the age of geocities and angelfire, before the great googleing..." $\endgroup$ Jul 7, 2016 at 16:42
  • 2
    $\begingroup$ I certainly interpret all-caps as shouting. I can't read it any other way. Since it's likely to be interpreted by many as rude, it's best avoided. $\endgroup$
    – Glen_b
    Feb 13, 2019 at 0:19
  • 1
    $\begingroup$ TROMP TROMP TROMP is too close for comfort... $\endgroup$
    – Nick Cox
    Jan 28, 2022 at 18:51
23
$\begingroup$

It begins "How do I use SPSS to..."

I was advised to do [procedure X] in SPSS. Here's a dump of the output. How do I interpret it?

Yes, I am singling out SPSS because it is prominently associated with bad questions. Of course there are thoughtful SPSS users out there, newbies and students included. Please formulate your question in a way that reflects that!

$\endgroup$
1
  • 14
    $\begingroup$ And a lot of our Python questions are to machine learning what SPSS is to statistics. $\endgroup$
    – Sycorax Mod
    Nov 1, 2016 at 13:47
22
$\begingroup$

A post that shows no concern for its readers does not deserve answers.

  1. Teh speling en gramar r 2 terse and so ful of misstakes that we can scarsely reed the queston. Its ovbious teh poster hasnt even red it theyselfs!

We should be sensitive to people who are not native writers of English and work to accommodate them. Even so, there's a fairly clear line between struggling with the language and being completely careless. When the language in a post is more mangled than even a free Web translator will produce, expect people to vote to close it.

$\endgroup$
1
  • $\begingroup$ You are right! SPSS is like from 20 years ago. The hot questions now are "How do I fit my neural network using Python and sklearn?" $\endgroup$
    – David
    Jul 5, 2019 at 10:43
21
$\begingroup$

[We often don't close these - but I increasingly tend to think we should put them on hold until clarified]

The title is:

What test should I run?

but which contain no indication of any question of interest whatever, instead focusing entirely on a description of the data (and that usually omitting important information), as if the form of the data were the only relevant factor in deciding what tests are suitable. [The use of the word "run" rather than some other verb often acts as a hint that the asker is using SPSS.]

I blame the way stats is often taught and especially the tendency to focus on following a recipe. In particular I think these questions tend to follow from --

  • an uncritical focus on Stevens' typology (nominal/ordinal/interval/ratio) as if that were the only factor of relevance (or indeed the only way of categorizing variables).

  • prescribing some kinds of analysis and proscribing others with reference only to that typology rather than to the substantive questions of interest and to the properties of the inferential procedures considered. Many times an analysis that would be proscribed in this way (or simply not considered at all) turns out to be a perfectly sensible as a way addressing the questions of interest, while the prescribed analysis may sometimes be considerably less useful (or even misleading).

  • over-reliance on a "cheat sheet" or "summary table" or other recipe that lists an analysis to go with a form of data, which often blocks people from searching for anything better.

These questions are also associated with a common tendency to think that parametric procedures are limited to normal-theory procedures and nonparametric procedures are what you use when you reject a test of normality; often there's some perceived issue with normality that prompts the question.

Besides, the publishing house O$\hspace 0.2mm$RLY$\hspace{0.1mm} ^?$ has an entire volume dedicated to the subject.

Image of the front cover of a book in the style of an O'Reilly book, which has the title "What Test Should I run?" with subtitle "A vaguely definitive guide". There's a large black and white drawing of a bat in a very old style. The author is Noah deCernibal-Hypothesis and there's a small line right at the top that says "Asking the unanswerable question"

(click the link or the picture for a larger image)

$\endgroup$
10
  • 1
    $\begingroup$ I've never heard about this "typology" before so had to google - and it looks like it's Steven's, not Stephen's. $\endgroup$
    – amoeba
    Nov 24, 2016 at 16:19
  • $\begingroup$ @amoeba Ah, thanks, I'm always doing that. I meant to double check before clicking post. $\endgroup$
    – Glen_b
    Nov 24, 2016 at 16:33
  • 5
    $\begingroup$ @amoeba you have no idea how lucky you have been to have avoided it all this time. $\endgroup$
    – mdewey
    Nov 24, 2016 at 16:45
  • $\begingroup$ @mdewey Can you elaborate? I am curious. If "Steven's typology" is simply distinguishing between nominal/ordinal/interval variables, then what's wrong with it? These are indeed different. I am not quite sure why this elementary distinction deserves a name. $\endgroup$
    – amoeba
    Nov 24, 2016 at 22:08
  • 1
    $\begingroup$ @amoeba because I agree with Glen_b that many of its disciples have an uncritical acceptance of the view that it completely determines what they should do with their problem. At least, that is what I think he said in part. $\endgroup$
    – mdewey
    Nov 24, 2016 at 22:12
  • $\begingroup$ @mdewey Indeed it is. A common problem is what it leads people to do with 0-1 responses. $\endgroup$
    – Glen_b
    Nov 25, 2016 at 0:50
  • $\begingroup$ The image that was put in in an edit (thanks @Sycorax) was a bit large. I have converted to a link for now but I may come back and figure out how to make it display smaller $\endgroup$
    – Glen_b
    Apr 11, 2017 at 23:29
  • $\begingroup$ @Glen_b meta.stackexchange.com/questions/25051/… I think this is what you're looking for. $\endgroup$
    – Sycorax Mod
    Apr 11, 2017 at 23:37
  • $\begingroup$ @Sycorax Was already there, thanks (I asked some other mods). Just making sure all looked good before posting. $\endgroup$
    – Glen_b
    Apr 11, 2017 at 23:40
  • $\begingroup$ Cool. The picture is just too good for it to get lost to the chat history. $\endgroup$
    – Sycorax Mod
    Apr 12, 2017 at 0:22
18
$\begingroup$

Duplicates or near duplicates

This site aims to collect questions and high-quality answers that are useful not only for the OP, but also for other readers who face the same question. This site exists for over 5 years and has seen something like 83,000 questions (as of July 2016).

So if you have a question that, for example, relates to a key topic of an introductory or intermediate statistics class, there is a very good chance that somebody else already had a similar question on this site.

It is therefore good practice to do a little search on this site before asking your own question. This has advantages for everybody:

  • It is time-saving for those users who donate their time to answer questions and moderate this site.
  • You may benefit from excellent answers that already exist, with no delay.
  • It helps keep the site uncluttered.
$\endgroup$
3
17
$\begingroup$

Posting a question as an introductory foray into conversation

Hey guys!

Well it seems to me like like your doing a lot of test anxiety around here. LOL Idont really understand this stuff, but i want to learn! Anyway I was just wondering if my [vague ideas about stats],or really big data/machine learning (haha jk) are even [possibly to meaningfully articulate] here.

i dont know what do you think?

Thank you very much. Avery

Treating stats.stackexchange (or any other SE) as simply an online forum for discussion should prompt immediate closure. A kind approach would be to direct the author's attention to other sites (e.g., reddit, etc.) in a comment.

$\endgroup$
4
  • 2
    $\begingroup$ Unfair to Averys the world over. More seriously, the pattern here is to follow a generic poor question with explanation of why it is poor and if possible what to do better. $\endgroup$
    – Nick Cox
    Sep 20, 2017 at 21:33
  • 2
    $\begingroup$ @NickCox I confess I was tickled by the prospect of the opportunity to provide an object lesson as an example. I mean lolwut?! $\endgroup$
    – Alexis
    Sep 20, 2017 at 22:04
  • 1
    $\begingroup$ @NickCox And thank you for urging my compliance with the illustrative pattern. Like my edit? $\endgroup$
    – Alexis
    Sep 20, 2017 at 22:11
  • 1
    $\begingroup$ I haven't seen questions like this here, but a pre-emptive post does no harm. Know that I had to look up "lolwut". I have seen questions answered helpfully on Reddit; in my limited sample it's Quora in which the ratio of ill-informed opinion to logic and evidence can approach singularity. You missed "Thanks in advance" which drives old fogeys to apoplexy. $\endgroup$
    – Nick Cox
    Sep 21, 2017 at 9:23
16
$\begingroup$

How do I interpret this output?

This kind of question consists of copying and pasting some output from your software and then asking us how to interpret the output. (In really bad cases, the output isn't even formatted to respect line breaks or use of fixed width fonts and thus to echo the layout in your software, so it's essentially unreadable too.)

Why is this often a bad question?

What does "interpret" mean, any way?

Is it: Tell me what the results mean? Sometimes people want a substantive interpretation and are asking that someone writes a story for them using subject-matter knowledge, factual and theoretical. That is, it's economic data, and you want to know what a (student or beginning) economist should write to satisfy other economists. Or it's psychological or biological data or whatever; only the peer group differs, but it's the same expectation. That's usually a tough call even if experts in your field recognise the variables and could do it, but almost no-one on CV wants to help to write your term paper or project report or to draft your submission to a learned journal. We're happy to advise on specific, concrete statistical points.

Is it: I'm new to this technique and I want you to go through the output and tell me how to think about it? That's usually far too much to ask. Are we to assume that you know nothing and need everything explained? That's like asking us to write a lengthy personal tutorial, but those already exist. That's the lecture you should have had or the textbook chapter you should have read (or even the Wikipedia article that is, surprisingly, quite good). By all means, ask specific technical points, while showing that you have made some effort to understand what you are doing, but don't expect that how do I interpret this output? is a switch that will set an expert going on a detailed interpretation.

$\endgroup$
3
  • 3
    $\begingroup$ Your answer here has been referenced in a discussion about whether the interpretation of statistical output is on topic. My position is that in general it should be considered on topic; but I recognize that such questions could still be unclear or too broad & may be closed for those reasons. I take that to be your point here. I just don't see anything specific to output interpretation questions as necessarily making them any more or less likely to be unclear or too broad than questions on any other subject. $\endgroup$ Jun 22, 2018 at 20:45
  • 2
    $\begingroup$ The main suggestion here is that such a question may be a bad question. If people stop at the title here and don't read further, they will miss the explanation of why. I've made some small edits to move closer to your point. (I've sometimes quoted this answer when a question on CV is essentially nothing but an output dump, but conversely there is no need to quote it if there is a clear statistical question.) $\endgroup$
    – Nick Cox
    Jun 23, 2018 at 6:40
  • 1
    $\begingroup$ @gung Most of the answers in this thread are about asking questions the wrong way and flag elements that don't help a question while not necessarily being immediately fatal. For example, "I need help" usually flags a naive or unclear question. Sometimes the poster is so new to forums that it's not yet obvious that almost everyone asking needs help too. Sometimes the poster doesn't yet have a specific question. But a title starting "I need help" doesn't rule out a good question following. It can just be someone wanting to make their question human or opening a conversation. $\endgroup$
    – Nick Cox
    Jun 23, 2018 at 7:32
15
$\begingroup$

Can anybody help me?

The kind of question here often starts "I am an X working on Y" with perhaps a little more detail, and then ends with an open request for help.

Why is this a bad question? Isn't the point of CV to offer help to those who ask?

What's usually bad about this is that it confuses the site with a helpline. A helpline I'll define in the following way. It's someone's job on a helpline to be available, perhaps in a designated room, or at the end of a phone, or at an email address. If you ask for help, it's then their job to listen, very likely to ask further questions and get information from you, and provide what help they can. It's also their duty to be respectful, even deferential. Their role is always to try to help to the maximum extent (subject to various simple rules such as your being a customer or more generally entitled to help). This applies whether the person asking is (say) a customer, a student or an employee.

The helpline analogy really doesn't apply to Cross-Validated. It's no-one's job to answer your question, or even to signal their willingness to do so. That follows from the principle that no one need answer any questions at all.

The way the site works is more like this. It's your job to work hard at a good specific question. "I am working on Y" doesn't qualify. What do you want to know, concretely and precisely? Then if you did that well, it's likely that someone will want to answer. So the answer to "Can anybody help me?" is in essence that someone is likely to try, but only if you ask a good question.

$\endgroup$
13
$\begingroup$

I just invented a square wheel: is it a good idea?

We often have threads where someone claims to have just solved some basic problem (usually with an equally basic solution, like "you just need to add 42 to the output"). Those questions often get closed, but sometimes they get answered. The answer usually points out where the OP made a mistake that invalidated their brilliant solution.

The problems with such questions are that (a) they usually do not ask any valid question and (b) they focus on the solution rather than the problem, so it is hard to grasp what problem they are inclined to solve. Moreover, in many cases they would fit better in a discussion forum where they could be discussed, rather than a Q&A site where at best we can enumerate the reasons why the approach is right or wrong.

Such questions do not seem to fit a Q&A site unless the OP defines them in terms of a problem to be solved, rather than presenting a solution.

$\endgroup$
2
  • 1
    $\begingroup$ Is this similar to the XY problem? $\endgroup$ Nov 27, 2017 at 18:38
  • 4
    $\begingroup$ Although it's usually undesirable as a direct putdown -- as being neither courteous nor constructive -- the reaction "That's not a question; that's a statement" does often summarize posts of this kind (and some others). $\endgroup$
    – Nick Cox
    Nov 27, 2017 at 19:07
13
$\begingroup$

Questions that provide no information whatsoever.

My favorite of these (which I think was on another list) is:

"I have some data. Can I do a t-test?"

$\endgroup$
2
  • 4
    $\begingroup$ +1. We get those questions, too. $\endgroup$
    – whuber Mod
    Jun 25, 2018 at 12:27
  • 3
    $\begingroup$ @whuber that specific one is a stronger case, 'I want to do some statistical thingies with my data, what do I do?' $\endgroup$ Jun 25, 2018 at 15:16
9
$\begingroup$

Answer as soon as possible

When I see this in a question I have an urge to close the question even without reading further. I do read further and usually close the question, because in most cases those are homework assignments that someone wants us to solve for them because they didn't bother and the deadline is coming. Those are sometimes also not homework, but other problems, but usually "I need an answer ASAP" means that someone didn't plan ahead, or didn't think over the problem carefully and now they are desperate. In the latter case the description of the problem is usually chaotic and unclear. In both cases the meta answer is that the OP should plan ahead solving their problem: they could even post a question at an earlier stage and that would increase their chance of solving it on time. Moreover, this is not a question-answering service, so nobody should demand answers to their question "ASAP". I find that rude.

$\endgroup$
2
  • 6
    $\begingroup$ +1. When the post is particularly insistent that we reply, and do so immediately, I also downvote it, even (as is typical) the OP is new to the site. I strongly support making newcomers welcome, but I don't lose any sleep over carrying out this particular voting policy. It proactively improves the site, because when the same person repeats their attempt in another post--which frequently happens--a second downvote (by me or anyone else) will cause the system to block them from posting any new questions until they have improved their existing posts. $\endgroup$
    – whuber Mod
    Jun 23, 2020 at 12:24
  • $\begingroup$ Not just homework. I've seen it from people whose thesis is due in 48 hours or less - I've even seen it as short as six hours. These posters have (apparently) not even begun some crucial-to-the-thesis statistical analysis. This has cropped up not just once or twice but many times; I see this last minute "thesis data analysis" typically with "my supervisor went on vacation" question multiple times a year. How thoughtless of the supervisor to assume the student would begin their analysis before the last possible moment. $\endgroup$
    – Glen_b
    Dec 5, 2023 at 3:19
8
$\begingroup$

Data is included... as a screenshot (or scan)

data

We are often better able to help you if we know the data you have. However, a screenshot (or scan) of your data is usually useless. Please don't expect us to type your data off ourselves!

At the very least, paste your data into your question as actual text and numbers. If you use R, paste in the output of dput(your_data), which has the advantage of giving us the exact data you are using, including all the "invisible" attributes.

For instance:

> your_data <- ts(rnorm(10),frequency=4)
> dput(your_data)
structure(c(1.05933915992086, 0.0232705232306941, 0.559884365949286, 
-0.420069990155258, 0.779473916276636, -1.29044555426724, 0.881259350107899, 
1.42835014304511, 1.97922676401677, -0.0579681924819385), .Tsp = c(1, 
3.25, 4), class = "ts")

If you just pasted in a sequence of ten numbers, we would not know that this was a ts (time series) object with a frequency attribute set to 4, so probably quarterly data. This can make a difference! So your question should ideally contain the following:

your_data <- structure(c(1.05933915992086, 0.0232705232306941, 0.559884365949286, 
-0.420069990155258, 0.779473916276636, -1.29044555426724, 0.881259350107899, 
1.42835014304511, 1.97922676401677, -0.0579681924819385), .Tsp = c(1, 
3.25, 4), class = "ts")

This is much more useful to us than the screenshot at the beginning of this post.

If you make it easy for us to help you, you have a better chance of actually getting help.

$\endgroup$
2
  • 3
    $\begingroup$ "paste your data into your question as actual text and numbers": even if people don't use R, data in the form of your screenshot are manageable in many different programs, so long as they (a) can be copied and pasted as text (b) have clear, informative headers (c) are clearly separated into columns, meaning variables. $\endgroup$
    – Nick Cox
    Jul 3, 2020 at 17:28
  • 2
    $\begingroup$ "Please don't expect us to type your data off ourselves!" Oh. Totally. Yeah. I am sure none of use would do that. ;) $\endgroup$
    – Alexis
    Feb 13, 2022 at 23:36
7
$\begingroup$

Here's the entire train of thought I had from the moment I dreamt up the project till I just finished writing this question, including every irrelevant detail and conversation with my supervisor (who once taught a stats course and is thinking about writing a book, except they had a skiing accident and got sidetracked, but on the upside they now have some time to read my manuscript but think I did my analysis wrong), plus the flashes of questionable insight I got while pondering this in the shower this morning, though I forgot some of it because my allergies started kicking in, isn't it crazy how that happens just when you've got some important idea, but anyway what I wanted to ask is...


I get that it's sometimes hard to tell the difference between relevant and irrelevant detail when framing a question. But when I have to read through 4 meandering paragraphs before getting the tiniest inkling of what help is needed, I'm not inclined to react favourably. These questions are unclear because they distract with excess detail. And they are worse than some of the other types of bad questions because they waste more of the reader's time.

Bonus points for combining this with Can anybody help me? and What do I do with this data?.

$\endgroup$
2
  • 3
    $\begingroup$ +1, but add the positive advice: Clear and concise explanation always is a bonus. $\endgroup$
    – Nick Cox
    Jul 23, 2019 at 14:41
  • 1
    $\begingroup$ @NickCox Agreed, I try to be less cranky when commenting on the questions themselves. $\endgroup$
    – mkt
    Jul 23, 2019 at 15:15
6
$\begingroup$

Please check my homework

These questions post a routine exercise, give (often in great detail) an answer (that often is correct) and ask

Is my answer correct?

Why is this problematic?

  1. We are not a homework-checking site.

  2. The question effectively has been asked and answered. What is any answer supposed to add?

  3. In most cases, the question is a duplicate of another post, anyway.

How to improve such a question

Presumably, if you are posting such a question you have some doubt about what you are doing.

  • Isolate the step in your solution at which you first experience a concern.

  • Show us your work through that step.

  • Show us what steps you have taken to check your work up to this point.

  • Describe the research you did to find answers here on CV (and elsewhere).

  • Express your concern: what is it that causes you to think you might not be correct?

At this point you likely have a definite, constructive, answerable question--and the responses you get will be more helpful to you than a simple "yes, you got the right answer."

$\endgroup$
5
$\begingroup$

What is wrong with my code?

I am working on a statistical problem. Here are pages and pages of the code I wrote (followed by those pages). What am I doing wrong?

Why this is problematic

  1. It requires everyone to read through your code just to understand the problem. Did you comment it, by the way? Did you even tell us what language it is in?

  2. Your code evidently doesn't do what you want. Perforce it does not convey what you do want. You need to state your question in (English) words.

  3. The question reads like it's asking for coding help: it belongs on StackOverflow, not here on CV. But it likely won't be acceptable there, either: see their help and guidelines for asking good questions. Look especially for information about constructing a minimal reproducible example. This is why we cannot migrate the question.

How you can improve the question

  • Describe your problem in words (perhaps supplemented with commonly understood mathematical symbols).

  • Make it clear that you are not asking for programming help or debugging: describe the statistical issue where you need help or better understanding.

  • Strip any code to its bare essentials. Make it reproducible, small, and clear. State what you expect (or hope) it to do. Include (small) inputs we can use to reproduce the problem. Display relevant output.

$\endgroup$
3
$\begingroup$

Is it okay for me to delete outliers or other data I don't like?

This is a surprisingly popular question. It's usually phrased in terms of 'improving normality' or 'reducing noise' or 'improving model performance'.

Most of this stems from bad stats education, basic misunderstandings and a desire to apply a simple technique to data that is ill-suited to it.

As a rule, a data point should not be removed unless it is an error. Questions that do not explain in sufficient detail why their situation may be an exception to this rule should be closed because any answer would be reiterating a simple point that has been made over and over on this site.

There are several threads about this covering common ground, but the best duplicate target I could find seems to be Is it OK to remove outliers from data?

$\endgroup$
4
  • $\begingroup$ I'm not sure this topic belongs as an answer to this thread. I would delete this and ask a new question about the issue. Let people discuss what is / should be the canonical thread for the topic. $\endgroup$ Aug 24, 2022 at 11:27
  • $\begingroup$ @gung-ReinstateMonica I agree it's somewhat different from the rest. I posted this at whuber's suggestion: chat.stackexchange.com/transcript/message/61847310#61847310 . I could remove the dupe target suggestion if that's out of line. $\endgroup$
    – mkt
    Aug 24, 2022 at 11:29
  • $\begingroup$ Hmm, I hadn't noticed that it was a suggestion to answer here. I thought it was a suggestion to post the issue to Cross Validated Meta, like @Sycorax did here. $\endgroup$ Aug 24, 2022 at 14:04
  • $\begingroup$ Cultures differ. In dealing with some kinds of economic data, it seems routine to Winsorize heavily before the main analysis. Then there are follow-up questions (seen by me in other places) e.g. Should I Winsorize before transformation or after? Do I need to Winsorize all my variables or just those with outliers? I often challenge people to cite a textbook or authoritative review explaining why such Winsorizing is a good idea, and have yet to receive any reference by way of reply. Winsorizing to get a Winsorized mean or variance I get, but not this as a good route to a robust model.. $\endgroup$
    – Nick Cox
    May 19, 2023 at 8:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .