Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Helping new users pose answerable questions

    Every time I pull up Statalist I see questions that are hard to answer because the user has not provided enough detail or failed to do other things suggested in the FAQ such as using code delimiters or dataex. Often, but not always, someone responds by pointing the questioner to question 12 of the FAQ and/or giving specific advice. I've done this numerous times, but it gets tedious and I often skip over the query hoping that someone else will do the job.

    Although one might think that first-time users of this list would click on the "Advice on Posting" link and read through the FAQ as a teacher I learned a long time ago that if you ask 100 students to follow a set of instructions and 50 of them do not the problem may be with the instructions rather than the recipients. I think that is the case here. The FAQ, well written, useful and informative as it is, may be a bit formidable, particularly to less sophisticated users. If you are eager to get an answer to your question, are you going to read through the 18 items in the FAQ? And are you going to realize that question 12 is probably the very thing you need?

    I wonder if there might be a way to program the list so that the first time someone posts a query they automatically get back a message that points them, with a link, to FAQ 12? The message could be gently worded to say something like,
    We are happy to see that you have joined Statalist. Although you may already be aware of it, we would like to direct you to our FAQ, particularly question 12, which provides suggestions and examples of how you might make your query more understandable and more likely to be answered.
    There might also be a second link that leads to a fictional exchange showing the use of code delimiters and other tools such as dataex. Of course, ideally, people should get this message before posting and one can imagine a barrier to posting until the user certifies that she or he has read the FAQ but I don't think we would want to do that.

    I know, I know, who has time to do all this? But who has time to repeatedly tell new users of this list how to use it most effectively?
    Richard T. Campbell
    Emeritus Professor of Biostatistics and Sociology
    University of Illinois at Chicago

  • #2
    Richard: Thanks for this. I've been notionally in charge of the FAQ for about 20 years, as you will know: these issues exercise me as much as anybody.

    What can I say that hasn't been said before?

    * There are plenty of conscientious members here who hit the ground running and post good code and data examples straight away.

    *. It is genuinely hard for many users to know what information is needed to make their question clear. If they really understood their problem, they could probably solve it, and no post would be needed. It's hard for many people to grasp the programmer's mindset that the details are crucial. They don't want to write code; they just want results!

    * The prompt when people (first posters, everyone) starts writing a new question enjoins them to read the FAQ Advice first. Some people are ignoring this every time. Some people even ignore explicit reminders.

    * The FAQ Advice is always longer than anyone wants to read, but it is divided into sections, and every few years we try to trim it down again. If there is any section that is not needed by someone somewhere, we need to know.

    * I have to guess that the main reason for not reading the FAQ Advice is mundane. Almost everyone feels major time pressure and will skip whatever doesn't seem essential. Not sweating the small stuff is life advice that many people follow. In programming and on Statalist, it is quite the wrong way round.

    I watch many more forums and social media than I participate in. Stata questions are posted in many places. Wherever the set-up is simple, Stata questions are usually futile because a good question needs more than 140 characters, or whatever, and so does a good answer. Whenever the set-up is more complicated, it is clear that many people struggle to understand how a site works. That is, for example, just as true of Stack Overflow and Cross Validated as it is of Statalist.

    Comment


    • #3
      I've had Richard's experience with those darn students being unable to follow my perfectly clear instructions, too. <grin>

      It's my casual impression here that the posters of questions without sufficient detail, seeming to blithely ignore the FAQ, are more likely to not be native speakers of English, which I take to arise from the even higher effort/time required to read the FAQ in a language other than one's best. One thought I've had is whether the FAQ could be made available in several languages. (Something like: "While your best luck here is likely to come from posting in English, we'd like to help you get started with an FAQ in several languages. Please choose the best one for you.")

      Perhaps volunteer translators for various languages could be recruited from among list participants? This is an utopian idea, I suppose, but the space and time devoted to clarifying questions has become distracting. (I must say, btw, that I'm frequently amazed at the ability of people on the list to decipher an unclear question.)

      Comment


      • #4
        While I agree that there are many on this Forum for whom English is a second language, and some for whom communicating in it is a real struggle, I don't have the impression that failure to follow recommended practices about posting is more common among them. (I haven't collected any data, this just an impression.) I do wonder if something like Dick Campbell's suggestion might be helpful. In addition to "pushing" a reminder to review the FAQ, I also suspect it would help if a YouTube video with demonstrations about using -dataex- and code delimiters were linked to the reminder. Even for native speakers, some people might find these things more easily understood with a video. StataCorp already has a number of YouTube videos demonstrating various aspects of using Stata. So they presumably have the facilities and know-how to produce one. Just a thought.

        Comment


        • #5
          Asking StataCorp for a video was also suggested to me privately. I will take this forward.

          Comment


          • #6
            In my opinion, a lot of the questions being asked since some time now, come from people who have only a vague idea of what they want to do -- either because they have been instructed to do something, have not understood the instructions and don't dare to ask for more detail --, or who couldn't be bothered reading up on the subject before posting, expecting a quick fix here. This is especially noticeable when deadline time is approaching In these cases,no amount of changes in the FAQ or videos would be of use.

            Comment


            • #7
              Art Kendall, who is well-known in SPSS discussion forums, has often remarked that questions posted to discussion forums (or posed to consultants) are rather like presenting problems patients bring to their physicians. In both contexts, a series of questions & answers may be required before one can get to the true underlying problem.
              --
              Bruce Weaver
              Email: [email protected]
              Web: http://sites.google.com/a/lakeheadu.ca/bweaver/
              Version: Stata/MP 18.0 (Windows)

              Comment


              • #8
                Bruce is right, and Art Kendall is too, but a forum like ours is not a helpline. That's the nub of the matter.

                I define a helpline as a triple

                1, Some people are entitled to help, as say customers, employees or students.

                2. It's someone's job to be willing to help. A questioner being confused, or uncertain, or not knowing is their problem too.

                3. The process of help can entail a lot of listening, a lot of extra questions, and so forth.

                The similarity with forums like ours is limited.

                Anybody can ask a question here, subject only to not posting spam, etc.,, but no-one has to answer anything and some questions go completely unanswered because it is hard to see what is expected that could be delivered in reasonable time. Certainly not complete personal tutorials on large topics or definitive suggestions for entire projects!

                We try quite hard to explain that, but strengthening the message would just lead to further flak.

                PS: I thought I should cheer people up with messages from some of our fans:

                https://www.econjobrumors.com/topic/...have-ever-seen

                https://twitter.com/simonmnaitram/st...86755063422977

                https://twitter.com/gabster0191/stat...905?lang=en-gb

                (The first and the second may be the same person. Multicollinearity bites again?)

                Comment


                • #9
                  Originally posted by Eric de Souza View Post
                  In my opinion, a lot of the questions being asked since some time now, come from people who have only a vague idea of what they want to do -- either because they have been instructed to do something, have not understood the instructions and don't dare to ask for more detail --, or who couldn't be bothered reading up on the subject before posting, expecting a quick fix here. This is especially noticeable when deadline time is approaching In these cases,no amount of changes in the FAQ or videos would be of use.
                  I was going to say something along these lines. I've been in a situation where I didn't quite understand what I was asking. For example, I had a bunch of data (Likert items for a scale, so ordinal data) with missing values, and I wanted a principled way to handle the missingness under the assumption of missing at random (conditional on covariates). I wanted to do multiple imputation and then calculate a polychoric correlation matrix. What I didn't understand was that the statistic in question does not have an asymptotic normal distribution, and hence it won't quite work with MI (although apparently there's some complicated transformation involving arctangent or some other strange trigonometric function that will make the R^2s asymptotically normal ... I didn't understand any of that explanation and I tuned out). Anyway, even among users of a fairly high ability level, the basic understanding may not be present when some sort of complex topic is involved.

                  After that thread, I definitely saw at least a few people who were doing something that they didn't quite seem to understand fully. Probably their professor told them to do X, and they tried, and they wound up having to come here.

                  And yes, there's nothing we can do about this. For my part, if I know what I am talking about, I try to help.
                  Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

                  When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

                  Comment


                  • #10
                    Update on #5: It's on Chuck Huber's to-do list.

                    Comment


                    • #11
                      Good morning to you all,
                      My name David, I am new to Stata and want Stata to estimate the threshold level but facing difficulties.
                      I try to generate new variable from my data but it keeps on giving error.
                      Please if you could permit me, i would like to attache my data and model for you to advice on what to do.
                      id year Volume Value Market Cap Inflation
                      1 2013 11.7 45,739,551 35408010 531.81
                      1 2014 15.5 79,472,245 71,419,962.81 553.74
                      1 2015 17.1 23,129,454 21,257,815.37 548.26
                      1 2016 17.5 11,835,752 10,946,399.90 411.2
                      1 2017 12.4 190,667,899 254,723,222.30 592.12
                      2 2013 11.7 14658999 65129460 1645.01
                      2 2014 15.5 6,776,134 46,640,857.62 2,228.54
                      2 2015 17.1 2 ,186,876 16,734,981.26 2,055.53
                      2 2016 17.5 1,308,656 8,924,413.42 1876.66
                      2 2017 12.4 3,396,754 24,642,645.00 2,228.54
                      3 2013 11.7 20780811 3701180 3032.97
                      3 2014 15.5 22,842,475 6,054,169.85 5,967.91
                      3 2015 17.1 13,623,044 4,123,181.46 6498.29
                      3 2016 17.5 13,206,588 2,622,574.05 2406.78
                      3 2017 12.4 15,227,384 2,055,929.71 3,850.84
                      4 2013 11.7 11946366 56389940 1285.25
                      4 2014 15.5 12,009,864 58,515,182.40 1,404.50
                      4 2015 17.1 9 ,634,831 41,723,786.82 1004.35
                      4 2016 17.5 8097974 28,571,555.65 943.4
                      4 2017 12.4 8,634,777 41,338,960.34 1,338.25
                      5 2013 11.7 117659419 71343680 1285.25
                      5 2014 15.5 5,476,004 6,152,659.29 445.2
                      5 2015 17.1 414,038 530,518.69 267.68
                      5 2016 17.5 629,298 551,335.79 223.07
                      5 2017 12.4 2,115,633 2,543,670.30 413.41
                      6 2013 11.7 0 0 0
                      6 2014 15.5 395,774 1,347,612.82 38.89
                      6 2015 17.1 168710 835,305.00 58.25
                      6 2016 17.5 0 0.00 59.69
                      6 2017 12.4 130,000 778,130.00 59.49
                      7 2013 11.7 885933 12893890 1725.68
                      7 2014 15.5 1,757,605 33,017,465.57 2,350.57
                      7 2015 17.1 1 ,283,802 24,011,453.03 12.41
                      7 2016 17.5 569,910 8,057,924.51 1406.88
                      7 2017 12.4 1,607,105 28,750,496.48 2,916.56
                      8 2013 11.7 49220 17476 70
                      8 2014 15.5 784,229 193,294.44 48
                      8 2015 17.1 1 ,280,656 332,138.98 74
                      8 2016 17.5 141,448 47,725.46 52
                      8 2017 12.4 360,100 114,205.00 70
                      9 2013 11.7 5138812 4118496 250.42
                      9 2014 15.5 9,968,245 8,958,211.41 333.89
                      9 2015 17.1 9 ,955,038 9,000,993.50 293.83
                      9 2016 17.5 2,341,363 1,714,884.15 266.02
                      9 2017 12.4 18,428,542 13,947,008.38 351.83
                      10 2013 11.7 13974994 6390647 205.34
                      10 2014 15.5 7,899,340 2,569,527.09 114.08
                      10 2015 17.1 17,245,602 2,583,259.59 45.63
                      10 2016 17.5 26,555,258 1,111,580.66 13.69
                      10 2017 12.4 17,054,641 613,869.79 0
                      11 2013 11.7 4105689 4265461 224.34
                      11 2014 15.5 6,507,487 6,216,011.51 264.83
                      11 2015 17.1 25,076,190 30,616,531.78 353.11
                      11 2016 17.5 27,733,165 41,344,797.30 431.05
                      11 2017 12.4 5,494,268 10,108,522.20 1,054.11
                      12 2013 11.7 10138234 1870617 19.04
                      12 2014 15.5 264,311 86,883.66 14.03
                      12 2015 17.1 943,781 149,235.72 9.52
                      12 2016 17.5 123,957 21,799.06 7.51
                      12 2017 12.4 541,611 43,961.35 3.01
                      13 2013 11.7 6328985 1512136 81.6
                      13 2014 15.5 7,623,375 1,270,396.25 57.6
                      13 2015 17.1 334,674 39,203.72 48
                      13 2016 17.5 1,510,352 80,967.24 28.8
                      13 2017 12.4 2,105,408 94,480.12 28.8
                      14 2013 11.7 4424263 35185175 566.08
                      14 2014 15.5 5,189,106 32,645,567.54 682.43
                      14 2015 17.1 1,037,626 6,189,509.16 570.56
                      14 2016 17.5 292,280 966,910.38 221.51
                      14 2017 12.4 10,766,101 25,274,735.47 394.92
                      15 2013 11.7 1927175 28216 769.29
                      15 2014 15.5 14,063.21 14,063.21 40.76
                      15 2015 17.1 3 ,522,874 67,105.19 40.76
                      15 2016 17.5 835,801 11,844.49 40.76
                      15 2017 12.4 472,371.00 10,191.52 40.76
                      16 2013 11.7 11126987 63640973 769.29
                      16 2014 15.5 4,759,617 14,063.21 610.09
                      16 2015 17.1 4 ,645,450 32,075,470.58 854.12
                      16 2016 17.5 10,390,246 89,225,924.70 1294.55
                      16 2017 12.4 4,254,828 58,636,765.63 2,056.87
                      17 2013 11.7 3588328 19852431 1310.3
                      17 2014 15.5 1,131,377 3,919,526.13 676.28
                      17 2015 17.1 1 ,502,780 3,678,071.01 854.12
                      17 2016 17.5 1,501,420 2,656,597.55 1294.55
                      17 2017 12.4 20,677,792 31,525,266.52 633.65
                      18 2013 11.7 41200 1648 1.36
                      18 2014 15.5 4,710 183.40 1.02
                      18 2015 17.1 15,236 457.08 1.02
                      18 2016 17.5 32,374 971.22 1.02
                      18 2017 12.4 2,600 78 1.02
                      19 2013 11.7 87000 3080 2.4
                      19 2014 15.5 4,200 126.00 2.4
                      19 2015 17.1 44,000 1310 2.4
                      19 2016 17.5 8,800 264 2.4
                      19 2017 12.4 0 0.00 2.4
                      20 2013 11.7 14210037 14076889 240.33
                      20 2014 15.5 12,653,117 21,390,589.90 224.92
                      20 2015 17.1 21,028,220 37,961,635.62 319.44
                      20 2016 17.5 7,072,552 17,227,566.43 319.44
                      20 2017 12.4 4,387,939 13,996,028.90 492.47
                      21 2013 11.7 11265527 4010990 76.3
                      21 2014 15.5 4,772,773 1,916,203.89 72.39
                      21 2015 17.1 14,594,679 3,111,256.98 27.39
                      21 2016 17.5 2,044,092 314,279.85 23.48
                      21 2017 12.4 6,157,030 667,429.66 19.56
                      22 2013 11.7 3900642 235098 2.19
                      22 2014 15.5 6,739,147 246,638.05 0.73
                      22 2015 17.1 1,325,961 16,519.22 0.73
                      22 2016 17.5 29,600 296 0.37
                      22 2017 12.4 4,891,610 48,916.10 0.37
                      23 2013 11.7 902156 100124 11.83
                      23 2014 15.5 1,700,327 68,898.48 4.73
                      23 2015 17.1 49,378,648 3,025,784.29 33.14
                      23 2016 17.5 1,062,575 133,185.32 33.14
                      23 2017 12.4 2,161,663 321,967.87 37.87
                      24 2013 11.7 902156 47370 36.55
                      24 2014 15.5 553,068 94,483.95 38.7
                      24 2015 17.1 7,374 1,319.17 38.7
                      24 2016 17.5 115,178,570 3,471,997.69 25.8
                      24 2017 12.4 95,780 9 ,584.00 21.5
                      25 2013 11.7 362467 128470 1.05
                      25 2014 15.5 105,720 12,973.20 38.7
                      25 2015 17.1 31,820 3,818.40 0.82
                      25 2016 17.5 25,100 3,012.00 0.82
                      25 2017 12.4 57800 6524 0.75
                      26 2013 11.7 2055245 309482 132.72
                      26 2014 15.5 1,120,905 498,778.00 50.4
                      26 2015 17.1 211,763 61,203.28 57.12
                      26 2016 17.5 154,480.00 35,320.17 36.96
                      26 2017 12.4 137,606 27,521.20 33.6
                      27 2013 11.7 3837372 16926803 2.96
                      27 2014 15.5 388,858 11,514.34 5.19
                      27 2015 17.1 1,362,900 41,494.00 10.39
                      27 2016 17.5 597,346 12,007.92 7.79
                      27 2017 12.4 978,300 21,460.00 7.79
                      28 2013 11.7 2471806 43504032 1144.38
                      28 2014 15.5 476,891 7,136,799.48 668.75
                      28 2015 17.1 109,002 879,792.66 531.25
                      28 2016 17.5 2,535,254 21,569,355.32 531.88
                      28 2017 12.4 618,894 6 ,166,252.09 802.5
                      29 2013 11.7 63040 2036292 31725.6
                      29 2014 15.5 78,482 2,605,100.93 31,734.67
                      29 2015 17.1 270 8,966.65 8,966.65
                      29 2016 17.5 11,159 302,619.06 24379.19
                      29 2017 12.4 9,689 170,372.11 23,562.21
                      30 2013 11.7 21836 11355 50.9
                      30 2014 15.5 281,020 146,655.78 50.9
                      30 2015 17.1 100 52 50.9
                      30 2016 17.5 112,656 57,909.14 50.9
                      30 2017 12.4 166,669 1,927.41 46.99
                      31 2013 11.7 33000 792 0.65
                      31 2014 15.5 3,928,800 121,702.00 0.87
                      31 2015 17.1 173,552 1,404,150.84 0.87
                      31 2016 17.5 19,984 798.49 0.87
                      31 2017 12.4 6,116 270.64 1.09

                      Yit =µi +β1×it h (qit < γ1) +𝛽2×it h (γ1<qitγ2) +𝛽3×it h (qitγ2) + 𝜀it

                      Thank you.

                      Comment


                      • #12
                        Anyone interested in #11 should please go to https://www.statalist.org/forums/for...hold-regressio

                        Comment


                        • #13
                          It is ironic to put question #11 in this thread.

                          Comment


                          • #14
                            HeIlo everyone, am doing a research, my data is a balanced panel, 7 years and 845 observations, I use the GMM, and the command xtdpd, but I want to control the effects of year and industry, when the command I enter the following, it gives an error, thank you for your help.

                            xtdpd etr L.etr ceonarcissismremuneration lev roa owncon duality fsize bsize ceoage bmeeting, dgmmiv(ceonarcissismremuneration lev roa owncon duality fsize bsize ceoage bmeetin) dgmm(etr, lagrange(2 4)) lgmmiv(etr) twostep artests(2)

                            Comment


                            • #15
                              Originally posted by Foad Zamani
                              I am doing a research, my data is a balanced panel 7 years and 845 observations, I use the GMM, and the command xtdpd, but I want to control the effects of year and industry, when the command I enter the following, it gives an error, thank you for your help.

                              xtdpd etr L.etr ceonarcissismremuneration lev roa owncon duality fsize bsize ceoage bmeeting, dgmmiv(ceonarcissismremuneration lev roa owncon duality fsize bsize ceoage bmeetin i.industry i.years) dgmm(etr, lagrange(2 4)) lgmmiv(etr) twostep artests(2)
                              Please post this as a new question instead of hijacking an old and unrelated thread.

                              Also, please see the FAQ to post a usable data example using dataex.

                              Comment

                              Working...
                              X