Added: 1 year ago
From: statslectures
Views: 9,661
Sort by time | Sort by thread (beta)

Link to this comment:

Share to:

All Comments (22)

Sign In or Sign Up now to post a comment!
  • Comment removed

  • There is at least one mistake in this video - ranks should be assigned based on the ABSOLUTE VALUE of the difference, NOT the signed difference. So 12 should be rank 1, -14 and 14 should both be ranked 2.5, 16 is ranked 4, etc.

  • Comment removed

  • @colman77

    It really depends on who you ask and what your professor or teacher tells you to do, unfortunately.

  • @statslectures

    Take a look at the Wikipedia page; especially the External Link to cornell.edu. It includes a great explanation for why this test works.

    with absolute value - this test approximates the probability the a dataset is centered around your test mean value.

    without - ... tests for strange cases depending on whether len(data) is even or odd.

    There are ongoing debates about a number of topics in statistics but I don't think this is one of them. Professors occasionally make mistakes.

  • @colman77

    I'll give it another look. I don't perform this test very often, and the way I did it here may be based on a flawed textbook. Thanks for letting me know.

  • @statslectures yeah it's a strange but interesting little corner of statistics. Thanks for taking another look, and thanks for the great videos!

  • My book says to calculate the Wilcoxon test statistic, use the SUM OF POSITIVE RANKS.

    I'm lost.

  • @AuronNU

    You use either the sum of the positive ranks, or the sum of the negative ranks, whichever is smaller. Maybe you're looking at an example from the book where the positive ranks were smaller?

  • @statslectures No it really states "The sum of positive ranks" and in the example the positive ranks are higher, in another example the positive ranks are lower, in both examples they take the sum of positive ranks.

    They test a Median agains a Benchmark, does that matter?

  • how should I perform Wilcoxon's test for one sample..eg: one sample data compared to a hypothesized median value..? Thanks in advance

  • first of all thank you very much for the video, very useful. I have a question, I tried to do your test with SPSS but it is giving me Z= -1.682 (Analyse>Nonparametric test>legacy Dialogs>2 related samples). can you please tell me what I am missing?

  • Very good and clear video. would have been nice if it was shown in a commonly used progem such as Excel as i got into a bit of a fix when trying to put that formula in excel. How do you get specific P-values from the Z-value, is that in an earlier video?

  • @goldtoleadalchemist

    Thanks. In the future I hope to make videos for statistical programs such as Excel, SPSS, SAS, and R. I find the specific P-values by looking them up in a Z-table, which you see around the 1:00 mark.

  • Hi again, i posted a question in your page. I have scores from two different systems, e.g. system I, graded 3, 4, 3. and system II graded 2, 4, 2. I want to compare if two systems correlates well. Wilcoxon Signed-Ranks Test seems a good fit after I understood, since its taking the ranking not the values.. What do you think? Thanks in advance.

  • @naushadzaman

    The test in this video is actually to see if the two groups differ. If you want to see if they're correlated, you want to use a different kind of test. If you're correlating regular quantitative data you usually want to use a Pearson's r correlation, while if you're dealing with ordinal (ranked) data you usually want to use a Spearman's r correlation. I'd have to know more about the data to choose the correct test, but check out those two videos first.

  • Hi again, i have another question, hope you don't mind. I have scores from two different systems, e.g. system I, graded 3, 4, 3. and system II graded 2, 4, 2. I want to compare if two systems correlates well. More specially, the second score was higher in both cases and first and third was same.. So they correlates well, even though the differences are not equal. If there is another system with lower differences that should be ranked higher. Which test do you think will be appropriate..

  • very good work.. can you please refer to the video where you explained how to get the value from Z table? 

  • @naushadzaman

    There should be a pretty good example of me using the z-Table in this video:

    "One Sample z-Test"

    You should be able to find it from my YouTube channel (YouTube isn't letting me link it directly).

  • @statslectures Hi. Heartiest thanks. Found it.. It was very useful.

Loading...
Alert icon
0 / 00Unsaved Playlist Return to active list
    1. Your queue is empty. Add videos to your queue using this button:
      or sign in to load a different list.
    Loading...Loading...Saving...
    • Clear all videos from this list
    • Learn more