Two-way repeated measures ANOVA in SPSS: one-within, one-between (March 2020)
Вставка
- Опубліковано 9 лип 2024
- This video walks you through a demonstration of a two-way repeated measures ANOVA where you have a single within-subjects factor and a single between-subjects factor using SPSS.
You can download a copy of the data here:
drive.google.com/file/d/15rOX...
Download the Powerpoint here:
drive.google.com/open?id=1Rfe...
Updated presentation using SPSS version 29: • One between, one withi...
My previous video on one-way repeated measures ANOVA can be accessed here: • One-way repeated measu...
Seriously. The best two-way repeated measure tutorial on the planet earth. Thank you!
You are a saviour sir. From last two days I was stuck with chat gpt and bird in inderstanding this. Your 20 minutes explanation solved it all.
Thank you does not cut it in explaining how grateful I am to you for this video. You have solved a month-old challenge for me. You provided all the answers in clear, simple and straightforward words. I could use it directly to interpret the data I am currently working on. I will definitely share it with my circle.
Your powerpoint saved my life thank you very much! Extraordinary help
This is so useful and everything is just so clear! Thank you so much!!!
Very helpful! Many thanks for this.
Thank you soooo much! Very useful! and sooo well explained! Many many thanks, Doctor Crowson.
Your explanation helped me to further explore my data for my doctoral dissertation from a different perspective. Especially the programming part at the end of comparing time and treatment groups. A simple "Thank You" feels inadequate as this little nugget of a video was a tremendous help! A heartfelt appreciation, for you Dr. Crowson!
You are very welcome, Todd. Thank you for letting me know. I wish you the best on your dissertation!
Thank you!! You really save my day!!
Thanks a lot for this video!!
Thanks a lot for the wonderful presentation
thank you for your videos Sir. Very helpful!!
Hi Sir, Thanks a lot. This video made my day!
Thanks so much for this, super helpful!
You're so welcome! Cheers!
Thank you very much !!!
Superb! Thanks
Good day, Dr. Mike! In your example, the interaction is significant How about if it has no significant effect? Thank you!
Hi, thank you for the video. Would you still use the same analysis if your Time 1 was a baseline (meaning that anxiety is expected to be the same across groups at T1 and then differ at T2 and T3). Thanks!
Hi @Mike Crowson! Thanks so much for this massively helpful tutorial video. I was wondering if you might have a sample corresponding write-up that you would be willing to add to the end of the slides or share elsewhere. This video has helped to walk me through my own data a great deal, but I'm struggling to figure out whether to discuss the overall model and then how to situate that vis-a-vis any presentation of the tests of within-subjects effects and tests of within-subjects contrasts etc. I've got Warner's Applied Statistics manual but find myself getting a bit confused as I move across multiple sources and I thus wonder if comparing your write up of this data to the video presentation of the analysis might allow me to see where I've run myself into confusion about what each of these bits are telling me with my data. Thank you in advance for your time and consideration (as a fellow academic I can understand what it is like to teach your classes and to also feel like you are running an additional course with internet students).
Thanks a lot! Great explanation. Please, could you provide any reference for Greenhouse-Geisser/Huynh-Feldt rules of thumb?
thank you so much you explain everything in detail. do you have any example or power point about the situation if we have one between factor and two within factor in repeated measure to share please
Hi Mike, thank you for sharing this video. I'm wondering whether I can compare the differences between each treatment at the same time point. For example, I would like to know if treatment A has a significant difference on the time 1 when compared with the control group.
Hi! If I am planning to compare the effect of two independent variable (say: type of sweetener and daily intake frequency) on one dependent variable (antioxidant capacity), do I still need to have a negative control group? Thank you so much!
I’ve just cried happy tears watching this. So god damn stressful
Thank you, Dr, Mike. I have one confusion. Suppose I have similar data as you have shown here but my main concern is I want to report statistical differences within the group at each time point. After using the last syntax as per your instruction I can see the diferences among them but I want to make sure that under univariate test result with G-G correction I can report that result for F value for interaction between time and treatment group or something else.
Dear Mike, your tutorial is awesome. Things are all clearly explained. I have one quick question, as I am running analyses for my research. Using your data as an example, if I am only interested in testing say Time 1 across conditions. I still use the same method as you did to control for the within-subject effect, but I don't seem to be able to obtain the contrast that I want (e.g., time 1 or time 2 in different treatment conditions). Can you please give me some advice? Many thanks
hi, what analysis would you suggest to me if I have a questionnaire with 43 question or item representing 5 categories of behaviour and I need to understand the differences between gender, age, nationality and also I have classified it using a cluster analysis so I have clustered and want to check the differences between clusters too.
superb bro
Hello Dr. Crowson, my interaction effect between the dependent variable (stuttering) and the independent (therapy group) variable were significant but meanwhile my pairwise comparisons between the two therapy groups (at the table shown at 16:37) were not significant...
I'm having trouble interpreting this, what does it mean?
Such a useful video it was! This video just made me relieved of confusion. There is only one question left for me. If there were two dependent variables (like Anxiety and maladaptive schemas), should I do two-way repeated measures for each one separately?
I have the same question
thanks!
Thank you. The explanation is very clear. Can you give an example on reporting the results of the two-way repeated measures ANOVA and how to arrange the tables (or what tables to be included?) in the results and disucussion section of the thesis?
what if you want to analyse both anx and depr scores in the 3 timepoints with 1 btw subjects variable?
Thanks for the tutorial, what if you don't have measurements of all the patients at all the timepoints, For example, for patient A you measured at 1 month and 2 months, and for patient B you measured at 1 month and 6 months and for patient C you measured at 1 month, 2 months and 6 months. Is there a way to let spss still include all patients, when I run the analysis right now it only includes the patients that I measured at all timepoints. Thanks in advance :)
Hi, I am a little confused where where the terms Linear and Quadratic came when observing the Tests of Within-Subjects Contrasts. Can someone clarify this?
Thanks a lot for this video.
You considered that all variances are equal. But the descriptive statistics show that the Std Dev is around 4 for all combinations, except for treatment B, at time 1 and 3, where Std Dev equals 1.24 and 0.95 resp. How should I analyze this data set if I consider these Std Dev are not equal?
Thank you for your wonderful video. Regarding the Levene test - I hope I understood you correctly - is it acceptable to "ignore" a positive test regarding one or two measurement points if the groups are equivalent (n largest / n smallest < 1.5) ? Do you have a source of literature on this topic?
In my research I have two groups and in each group I have taken measurements of pressure at 3 intervals ie 0, 2, 4, 8 weeks. I want to compare significance between 2 groups. Which test to use? How to say which treatment is better than other?
TOTAL ###WIZARD .. outstanding video, thank you Sir. Dumb question(s): if you only had two time variables (vs. three), like just a pre-post, would anything change? Is this different than a mixed-factorial ANOVA? Thank you.
Thank you!! I learned so much!! But my question is what is name difference of "A: Two-way repeated measures ANOVA" and "B: Two-factor repeated measures ANOVA", My teacher said this design is "B" not "A", so I am confused...Could you explain name difference of A and B?
It seems that in this video the Levene's and Box test assumptions aren't met in exactly the same way as in the two-way repeated measures I'm working on. I wanted to know what you would recommend? My groups are barely failing Levene's test (p=.039 and p=.049) but severely violating the Box test (p
Hi Deborah, Box's test is used if you plan to interpret the multivariate test results. If you plan to go with the univariate tests (e.g., sphericity assumed, Greenhouse-Geisser), then Box's test is not something you have to worry about (but of course, the assessment of sphericity would be with the univariate tests). The multivariate test results are useful if you don't want to worry about the issue of sphericity.
My opinion is that Levene's test results are largely irrelevant in this design, since the test of between-subjects effects is essentially an ANOVA testing the differences between the repeated measure averaged over the time points. This is usually nothing anyone really wants to test. So unless that is something you are interested in testing, then there's no real need to spend time worrying about Levene's test. [I cover Levene's test in the powerpoint for folks to know what it is. But I have generally found little use for it and the ANOVA testing between subjects effects.]
If you are still concerned with Box's test (because you want to focus your interpretation on the multivariate tests), yes one option might be to delete data. However, that would result in a huge data loss for you. One thing you might do is to consider the likely effect of a violation of Box's test on the multivariate test in terms of effects on power and/or type 1 error rate (I talk some about that in the Powerpoint) and discuss your results with the caveat of how power or type 1 error might have been impacted during your analysis. Of course, if you find significance when the test is made more conservative, then it's a "win" for you. I'll be honest, I have not spend a huge amount of time on different ways to assess the impact of a violation on power and type 1 error rate. So you might have to do some research on that.
But my general preference (in my own work) is to focus interpretation on the tests of within-subjects contrasts (univariate approach).
I hope this is helpful. Cheers.
@@mikecrowson2462 I appreciate the detailed response. It is helpful!
Thanks for going through the output table by table. I hate those videos that just say "we don't need to talk about it and it is not important", that just shows a lack of understanding by the video creator.
I'm so frustrated that SPSS Version 27 has changed this - and now I have NO idea on how to do a one-way repeated measures ANOVA.
Hi Teresa. I'm not sure what you mean regarding SPSS changing things. The one-way and two-way repeated measures ANOVA still involve the same approaches to running the analyses as in previous versions. I have a video on one-way repeated measures anova at ua-cam.com/video/TH7FVKevAcQ/v-deo.html . It was made on SPSS 26, but it still follows the same process. Also, I haven't put together a video on this yet, but you are welcome to check out a powerpoint I put together for a class on repeated measures ANOVA with SPSS 27: drive.google.com/file/d/1ZiNThxUyMMslPXdsHONli-nNhxja2z8o/view . The data is linked in the powerpoint.
@@mikecrowson2462 Thank you so much for getting back to me - maybe I have a weird version of SPSS 27, but when you go to Analyse, General Linear Model... The only option that now comes up is Univariate. Nothing else comes up. So it's different to SPSS 26 in that sense. Like I said, maybe I have a dud version. But that doesn't make sense either (but it could be that it's 3:30am here...)
@@teresamulhern Hi Teresa. One possibility is that the version you have is the 'base version' and not the 'standard version' of SPSS. I've had this issue arise with my own students. The base version if 'bare bones'. I went to the IBM website (www.ibm.com/products/spss-statistics/details) and see that to get repeated measures anova, you have to have the standard version (which includes the base version + Custom tables and advanced statistics. If you just purchased the base version, then you won't have that functionality. It's irritating that there are so many different versions of SPSS, and it's confusing to students. By the way, just an FYI, there is a free program called jamovi you can download here: www.jamovi.org/download.html . It would have the capability for repeated measures anova (but I haven't tried it out yet). Might tide you over while getting SPSS straightened out. Cheers!
@@mikecrowson2462 thank you so much for your reply - I really do appreciate it. Very frustrating as this is what is afforded by my workplace (and I'm meant to give a class on this soon to arts students). I'll figure it out!