Hi, thanks for your questions. The U-value measures the difference between the groups’ rank sums. It’s based on comparing each observation from one group with each observation from the other group, counting how often observations from one group are ranked higher than those from the other. Regarding the mean rank, it’s the average of the ranks for each group. Reporting it can help to illustrate which group has higher or lower scores on average.

Thanks Karen, glad it helped!

Hi, I'll add that to my list!

Hi, glad the videos have been helpful! If the p value for the Shapiro-Wilk test is below .05 (e.g., .032), it suggests that the data are not normally distributed. If you find that the variable representing the differences between the pre and post scores is not normally distributed, you could consider doing a non-parametric test instead (e.g., a Wilcoxon test).

@@DavidRobinsonPhD Honestly, thank you so much! Have a lovely week ahead of you!

@@math6661 Thanks, you too!

Thanks, they're on my list!

Thanks, I'm working on it!

Thanks, glad it helped!

Thanks, glad it helped!

Thanks, glad it helped!

It depends which are violated. Is there a particular assumption you're interested in?

Thanks Siobhan, glad it helped!

Hi Gemma, thanks for your question. It might relate to the size of the differences. For example, anxiety could be much higher in mums of autistic children than in mums of non-autistic children (hence the significant difference), and it could be only slightly higher in dads of non-autistic children compared to mums of non-autistic children (hence the non-significant difference). In this case, the difference between anxiety in mums of autistic children and dads of non-autistic children would be smaller than the difference between anxiety in mums of autistic children and mums of non-autistic children, which could explain why the differences between these pairs were significant and non-significant, respectively. You would have to look at, e.g., the medians to see whether this is a possible explanation. If not, yes, as you suggest, it could just be that the test has less statistical power when focused on the dads as there are fewer of them.

@@DavidRobinsonPhD Thank you so much this does confirm what I was thinking. Median anxiety for female caregivers of autistic childre n= 32.5, 11.25 = male caregivers of neurotypical children, 7.5= female caregivers of neurotypical children, score out of 100. (higher the number greater the anxiety)

Glad it helped!

Hi Mariano, thanks for your question. There's an explanation of this here: stats.oarc.ucla.edu/other/mult-pkg/faq/general/faq-why-is-the-mann-whitney-significant-when-the-medians-are-equal/

Yes, usually, values below 0.05 are considered significant. Therefore, a value of less than 0.001 would normally be considered significant.

@@DavidRobinsonPhD omg, thanks a lot sir, i am not confusesd again. Thanks sir

Thanks, glad it helped!

Hi, not yet, though I'm planning to make some videos with Jamovi soon.

Thanks Mariano, glad it helped!

Thanks for the suggestion. I've added that to my list!

@@DavidRobinsonPhD Thank you so much

Thank you for your informative video. Could you please suggest the correct method for performing ROC curve analysis when combining multiple parameters using SPSS?

Thanks Dom! Unfortunately, I haven't created one yet, though will add that to the list.

Thanks for watching!

@@DavidRobinsonPhD your channel is a huge help. If you're ever in Iceland I'll buy you a coffee. :)

@@sirrysif Sounds good!

Thanks!

Hi, thanks for your question! If you're following APA guidelines, while p values below .001 are reported as p < .001, other values are reported as they appear in SPSS. E.g., a value of .584 would be reported as p = .584.

​@@DavidRobinsonPhD Thank you. So can I say there was no significant correlation found rs = 0.042, n =170, p = 0.584.

@@user-tl8uv4co3b Yes, something like that would work. However, since making this video, I've realised it might be better to present this type of info a little differently: rs(170) = .042, p = .584. Note that in APA style numbers that can't exceed 1, like correlation coeffecients and p values, don't require a 0 before the decimal point.

Thanks Elnara, glad the example helped!

Hi Chef, if the items all measure the same constructs, it would be more common to complete the analysis on a variable that represents the sum or the mean of the items. Regarding your other question, yes, you could do a post-hoc test (e.g., Dunn's test) as the Kruskal-Wallis test just tells you whether there's a significant effect of the IV on the DV, not which specific groups differ significantly from each other.

@@DavidRobinsonPhD As i said, ,y items are formative in nature, it is not a reflective construct, so how can i combine them to a mean score for the analysis?

Thanks!

Thanks!

Thank you!

Hi, yes, ANCOVAs are often used to do just that.

@@DavidRobinsonPhD thank you very much Dr

Thanks, glad it helped!

Thanks Tracy, glad it helped! Yes, I also have videos on paired (dependent) and unpaired (independent) t-tests: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-2Gij9hzVtf8.html&ab_channel=DavidRobinson%2CPhD ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-PTYLCTNtOdQ.html&ab_channel=DavidRobinson%2CPhD

Thanks, glad it helped!

In APA style, it's the exact p-value (i.e., with an equals symbol): apastyle.apa.org/instructional-aids/numbers-statistics-guide.pdf

Thanks Rob!

Thanks Jon, glad it helped!

typing error : Backward LR

Yes, often in reports for university classes, the marker will expect to see, e.g., "(Appendix A)" in the results section as well as an actual appendix with the SPSS output so that they can check that you've run the analysis and have interpreted it correctly.