Kruskal Wallis non parametric one way ANOVA and post hoc tests (SPSS demo)

This video demonstrates how to carry out the Kruskal-Wallis one-way ANOVA using SPSS. The demonstration begins with a standard parametric one-way ANOVA and then moves to the demonstration of the K-W test using two menu routes. The second route provides an easy way to obtain pairwise tests of group differences (with those tests being Mann-Whitney tests).
You can download a copy of the data here to follow along: drive.google.com/file/d/1uuyg...
NEWEST VIDEO (RECORDED 2021) ON K-W TEST CAN BE FOUND HERE: • How to perform a Krusk...
For more instructional videos and other materials on various statistics topics, be sure to my webpages at the links below:
Introductory statistics:
sites.google.com/view/statist...
Multivariate statistics:
sites.google.com/view/statist...

[@georgiamayjay]

Holy cow, I just learned some major SPSS secret steps. Thank you.

[@coolingsounds5228]

The only KZbinr that makes sense😭you’ve saved my life

[@elizabethharris6427]

Thank you very much for this tutorial, has helped me a lot on one of my lab reports for my master's!

[@Wongerful]

On a rating of 1-10. You got 10. Thank you so very much for this video.

[@halabenghasheer2286]

Thank you very much. A clear and easy demonstration ..very useful

[@VaibhavYawalkar]

Thanks a lot. This is the best video on this topic

[@sanamajid1618]

Excellent tutorial! Thank you so much!

[@auyiee7244]

I have watched and followed every step in this video. It has been very helpful. Thank you so much

[@mikecrowson2462]

Hi there. I'm very happy to hear this was helpful to you. By the way, I have a much newer video (kzbin.info/www/bejne/b3bJnoqJYr51erM) from Feb 2021 that I developed on this topic that contains supplemental files you can download by following the links underneath the video description. Best wishes!

[@anaisjanik8201]

Thank you very much! Helped with my thesis!

[@visnjabartolovic9228]

Thank you, great video, very helpful!

[@mikecrowson2462]

SEE NEW VIDEO ON THIS TOPIC USING SPSS VERSION 26: kzbin.info/www/bejne/b3bJnoqJYr51erM .

[@atakanbilgili4373]

Exactly what i need, thank u so much foe clear explanation.

[@justrainmath8881]

This is by far the most comprehensive tutorial i saw. Thank you.

[@yra]

WOWEEE thank you very much!!

[@mikecrowson2462]

You are very welcome!

[@MLCHZDCK0518]

Thanks for making this easy to understand, how does the grouping of data work when using 18 years time series data with 1 and multiple independent variables? Alternatively how K-W test be performed on such set of data.

[@sshagiwal446]

Dear Mike, do you know how to activate the pairwise comparison function for KW in for MAC. I have SPSS v25 installed on my MAC but for some odd reason, I don't have the same functions as demonstrated in the video.

[@ahmadabedahmadi7520]

Thank you

[@ahmetgenc3141]

Thanks a lot. It is very helpful. One quick question. When I switched to pivot table & graph, I realized some of the not significant lines are bold (green) in pairwise connections. When I switch to model view, I could not see it in there. I tried to figure out the underlying reason from the table. Not successful. Any comment on why using bold lines? Thnx

[@redouaneilifi6378]

THANKS

[@irfankesumayadi8753]

Thank you, very nice video. Is that possible when we run Mann-Whitney test manually give different p-value compare to p-value from pairwise comparison ?

[@vinaamrina266]

Thank you so much sir

[@fatemehhadavand7505]

Hi, thanks for your helpful video, I do the same as you, but the software does not mention Bonferroni correction, what should I do?

[@qasimahmad6714]

When i try to do independent samples approach. Group section say it could not be continuous variable but when i used K independent Samples it does not have any issue with group. any suggestion?

[@anasuyadr5]

Hi pl, clarify. 2 groups with 3 different treatments? or one group with 3 different and considered as 3 subgroups?

[@slowmotion562]

Kruskal-Wallis* - but great video! Usually you specify x² and p ... The post hoc method in SPSS does not output an x² value, how do I deal with that?

[@chefberrypassionateresearcher]

Sir, please guide how to write the Kruskal wallis test and post hoc test results in thesis?

[@andreamatos3850]

Hello, thanks for you explanation, but I have a question: how do you ask or do for the pairwise comparisons of treatment, I do everything exactly like u and it doesn't show that.

[@gamalnasser5575]

Can the Bonferroni test be used after the Karuskal-Wallis test? Does the Bonferroni test require that the data depend on a normal distribution or not?

[@sanisacse6577]

How can I do Post hoc test?

[@user-xr1of2wx6p]

I want to ask something please. What does it mean if my null hypothesis is rejected but in the post hoc there are no adj sign values smaller than 0.05??

[@appusuriya6895]

nice, what is Bonferroni correction, please explain

[@umangsutariya7982]

Why the standard error is same across the table for each pair wise comparison

[@CH-is1vc]

Thank you. But I wonder, what is the name of the "pairwise comparison" that is default in SPSS as a post-hoc test? Is it Mann-Whitney or Dunn's test?

[@bilemiyorum_altan]

Dunn's Post Hoc. www.ibm.com/support/pages/can-spss-perform-dunns-nonparametric-comparison-post-hoc-testing-after-kruskal-wallis-test

[@Nur_9809]

You clicked on “All Pairwise” for conducting post hoc Kruskal Wallis, but how about if you clicked on “Stepwise stepdown”? What is the difference?

[@CarlosEObertoMachado]

Good morning, I see that the test name is written with the surname Wallace both here and on your blog. It is right? Or is it correct last name Wallis? Anyway, I appreciate your valuable input. Excellent explanation, very clear.

[@mikecrowson2462]

I believe it's Wallis. Just a misspelling. I plan on updating this video at some point with the correction :)

[@MsEmzRox]

When would you use the Bonferroni correlation?? For example my out put is looking at the difference between options no, yes and unsure, the sig. has differences between no and yes and I don't know and yes however the adj sig has no significance which I'm a bit confused about as the p-value for kruskal wallis is 0.02

[@mikecrowson2462]

Hi there. One typically uses the Bonferroni correction/adjustment when one is carrying out multiple tests as a way of keeping the Type 1 error rate under control (since the probability of committing a Type 1 error across a set of tests increases as the number of tests increase). The adjusted significance you are referring to are p-values that have incorporated this adjustment. The effect of using a Bonferroni correction is to make tests more conservative, so the p-values when using the Bonferroni-adjustment will generally be larger than those when it is not used. If you are finding non-significance using the Bonferroni-adjusted p's where you those p's are significant with the original test without the adjustment, that is simply the Bonferroni adjustment producing doing what it is intended to do: produce more conservative test results to reduce the probability of Type 1 error. Ultimately the decision about 'which way to go' is up to you in terms of whether you wish to use the correction or not. However, it does strike me that if your conclusions about which differences you have found differ between using the Bonferroni adjustment versus no adjustment, then you might go with the former. Again, it is ultimately your decision and one you should be prepared to defend. I hope this helps. Cheers!

[@ivonnesoeliono]

@@mikecrowson2462 Is there any reasonable excuse if we don't use the adjusted sig.?

[@ellemason2055]

Hi, what version of SPSS were you using in this video? Somehow, my SPSS version is not showing Chi-squared in the Test Statistics table. Instead it is showing Kruskal-Wallis test.

[@mikecrowson2462]

Hi there. This was probably SPSS version 24 or 25 when I put it together. I don't know about the differences across versions. However, just an FYI, have a much newer video on this topic (and you can download the data and powerpoint underneath the video) at this link: kzbin.info/www/bejne/b3bJnoqJYr51erM . I hope you check it out!

[@agnesberesinski3429]

Thank you very much for explanation, but I have a bit special case. In about 90% of my cases after the rejection of the null hypothesis one sees thanks to pairwise comparisons (post-hoc with Bonferroni corecture) which groups differ significantly. However, I have a few cases where KH rejects the null hypothesis (somewhat close but already 0.044) but pairwise comparisons show no significant differences. I thought I might look at the post-hoc values without Bonferorroni corectures and base the significance of the differences on them, because some of the significant values are already below 0.05. Is it possible to understand such an approach? Are the corrected significances according to Bonferroni possibly too conservative if KH is just below 0.05? Are there any alternatives ?

[@suradachengchoun1101]

Thank you a lot, why test statistic is negative value

[@mandaviupadhyay7016]

please dfenesion in hindi

[@febrip.s.p2016]

Thank you

[@mikecrowson2462]

You're very welcome. Thanks for visiting!