Have you ever used Wilcoxon signed-rank test, and if so, in what cases?
Currently, I analyze the results of copyright scheduling methodology, which aims to minimize makespan (for job-shop), and I intend to compare the data obtained using the methodology of the results achieved using the selected rule priority.
What are you experience with Wilcoxon signed-rank test?
Best regards,
Justyna
Wilcoxon signed-rank test is a standard test to use if you are comparing two groups of data that are not normally distributed. I did not completely understand what your data set is like, but if it consists of two groups of data (for example one set using this scheduling methodology, and another with some control condition) which you wish to compare, and the data is not normally distributed (basically it does not generate nice normal curves when you plot it) , then Wilcoxon test is a good choice. (You can even use it when the data IS normally distributed, it might not be the optimal solution, as a normal t-test could be more efficient).
I work in a Nursing School on Nursing Research Projects and I have used Wilcoxon-signed rank test on several different projects as a kind of nonparametric paired t-test. Usually, I work on instrumentation studies that involve composite scores computed from ordinal Likert items that are not always parametric for several reasons (sample size, how the scale works in a sample). As a result, I find it difficult to use a standard paired t-t test (difficult to explain in publication). So, I end up using the Wilcoxon in place of a paired samples t-test.
Dear Ilkka
Thank you for your answer.
In brief:
I generated 120 sets of production orders and I prepare two production schedule for each set - one using my methodology and second using another methodology. For both cases makespan was counted . I replaced the numerical values for percent values and made a 1-sample Wilcoxon test (single test population), generating a pair (x, m) where m = 1 (i.e. a value of 100% for my method). I proved in this way that my method minimizes the makespan (compared to the second method).
Dear Bruce
Thank you for your answer. It is very interesting. I have just preparing my first publication using Wilcoxon test so we'll see how I go to explain the text in the publication :) In my case it occurs large sample (n=120), so the statistical distribution tends to a normal distribution.
Best regards, Justyna
- Badges
- Science topic
- Similar topics
- Engineering
- Mechanical Engineering