+ Reply to Thread
Results 1 to 8 of 8

Thread: evaluating Kruskal-Wallis test results

  1. #1
    Points: 46, Level: 1
    Level completed: 92%, Points required for next Level: 4

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    evaluating Kruskal-Wallis test results




    Using an online statistical tool I evaluated 5 group of students' (self-identified ethnic groups with na= 105; nb= 50; nc= 44; nd= 43; ne= 33.) answer to Likert like question. The computer gave me the following answer:
    H=10.4
    Df=4
    P= 0.0398

    Am I correct in saying that the median for the groups are statistically different?
    Do I have to perform a Mann-whitney test on each combination?

    Thank you.

  2. #2
    TS Contributor
    Points: 12,227, Level: 72
    Level completed: 45%, Points required for next Level: 223
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,470
    Thanks
    160
    Thanked 332 Times in 312 Posts

    Re: evaluating Kruskal-Wallis test results

    Hi,
    if I understand correctly you had data like 105 a-s 50 b-s etc and the research question was whether you have about the same number of a-s, b-s ... with the alternative being that there is a systematic effect leading to more a-s, say, then would come from just random variation? If I am correct you probably should have used a chi-squared test instead of KW.

    regards

  3. #3
    Points: 46, Level: 1
    Level completed: 92%, Points required for next Level: 4

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: evaluating Kruskal-Wallis test results

    Thank you! The survey asked students to respond to questions e.g. “In my opinion, the economic system in the US leads to an unfair distribution of income” by indicating their opinion in a Likert scale. The 5 groups are their self-identified ethnic groups. If P= 0.0398 than does that signify that the median for the groups are statistically different? Do I have to perform a Mann-whitney test on each combination?

  4. #4
    TS Contributor
    Points: 40,715, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Downloads
    gianmarco's Avatar
    Location
    Italy
    Posts
    1,369
    Thanks
    232
    Thanked 301 Times in 225 Posts

    Re: evaluating Kruskal-Wallis test results

    The p-value indicates that at least one group significantly differs from the others. Usually, a follow-up test would be the Dunn test. Sometimes, some softwares perform MW test on a paiwise basis, with some sort of p-value correction (e.g., Bonferroni).
    As for the "median" issue, like MW test, KW is not a test of equality of medians. A broad interpretation can be thinking about the test as a means to assess if at least one group scores higher than the other ones. Since the test is based on ranks, maybe thinking of it in terms of mean ranks among groups would be also appropriate.

    Hope this helps
    http://cainarchaeology.weebly.com/

  5. #5
    Points: 46, Level: 1
    Level completed: 92%, Points required for next Level: 4

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: evaluating Kruskal-Wallis test results

    Thanks. I will explore the Dunn test.

  6. #6
    TS Contributor
    Points: 12,227, Level: 72
    Level completed: 45%, Points required for next Level: 223
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,470
    Thanks
    160
    Thanked 332 Times in 312 Posts

    Re: evaluating Kruskal-Wallis test results

    Btw,
    did you try a simple ANOVA? Only the residuals need be normal, so giving it a try could be worth it.

    regards

  7. #7
    Super Moderator
    Points: 13,151, Level: 74
    Level completed: 76%, Points required for next Level: 99
    Dragan's Avatar
    Location
    Illinois, US
    Posts
    2,014
    Thanks
    0
    Thanked 223 Times in 192 Posts

    Re: evaluating Kruskal-Wallis test results

    The post-hoc test for the K-W test is:

    \left | \bar{R_{i}}-\bar{R_{j}} \right |\geq z_{\frac{\alpha }{k\left ( k-1 \right )}}\sqrt{\frac{N\left ( N+1 \right )}{12}\left ( \frac{1}{n_{i}}+\frac{1}{n_{j}} \right )}

    Thus, with an alpha per-comparison rate of 0.05 and k = 5, the critical value of z from the standard normal distribution would be 2.80703.

  8. #8
    Points: 46, Level: 1
    Level completed: 92%, Points required for next Level: 4

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: evaluating Kruskal-Wallis test results


    Thank you, will try

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats