Jonckheere Trend Test - SPSS (part 2)

Although relatively unknown, the Jonckheere Trend Test can be quite useful, as it is sometimes significantly more powerful than the more common Kruskal-Wallis test. Effectively, the Jonckheere Trend Test (a.k.a., Jonckheere-Terpstra test) is the non-parametric analogue to the linear contrast analysis within the ANOVA framework. In this video, I demonstrate how the Jonckheere Trend Test is the only method to obtain a statistically significant effect, in comparison to the Kruskal-Wallis, Median test, and the One-way ANOVA. I also show how to estimate the effect size for a Jonckheere Trend Test via Kendall's tau-b. In fact, I show how Jonckheere's Test and Kendall's tau yield identical results.

[GHXL]

I just know that J-T test is actually THE SAME with Kendall Tau b test. I am in the middle of doing this on my dissertation when I saw your video. THANK YOU VERY MUCH Sir!!!!. I have subscribed as well.

[jessesaris]

Dear how2stats,
Thank you for the informative video, i've found it very helpful.
I have one question (I hope you still read these comments):
The trend I'm testing is not linear but has an exponential form;
when plotted, the increase of the Y-value grows smaller and smaller with the increase of my X-value (I hope I'm explaining this clearly)
Is there a way to test when the increase of the Y-value is no longer significant?
Thank you in advance for your response, keep up the good work!

[shailevi]

Hello, thanks for the video.
There's only one thing that I haven't understood yet:
I have to examine a possible trend of percentages across a period of 6 years (for example: 1995 to 2000) among two groups (in total: 12 percentage values. 6 values for each group).
If I will do "Split file", then,  does my "grouping variable" should be the year?

[swatithakkar]

hello all i have problem that how to decide the order  of the test i mean to say that we will do the K.W. test and if it is rejected than how to decide the oder say x1>x2>x3>x4 it its is our alternative so....???