Repeated measures ANOVA 2: A within- and between-subjects design

[@foadkhosrowshahi9118]

Hi, I've just seen your channel and noticed that you are from Iran. I'm Iranian and have been in the UK for around one year. I'm so glad to see your channel and CV and proud to be your compatriot! - Foad

[@nelsondelarosa5490]

I am simply grateful for your thorough and extremely helpful explanation.

[@scientific-accomplishment]

I learned so much!! Thank you!!

[@semiloreakande3620]

hey , how would you do this type of analysis if u have 2 between measure variables and just 1 within measure . thank you

[@imliyaibrahim9179]

Hi sir. Thank you for the great video. But I have one question. I have include few covariates for this analysis. And I'm aware there produce difference significant value for the (test within subject) effect and (estimated marginal means) for the same variable. Hence, I'm not sure, which result should I refer? Thank you so much sir.

[@RomanBuffalo]

Hi, the video was wonderful. Unfortunately I am trying exactly this in SPSS and having some difficulty understanding the results. If you have 4 different meds and anxiety levels at 4 different time with each med, SPSS is comparing mean of all 4 time points for meds A, B,C and D rather than the total decline in anxiety.. So if with A it went from 12 to 3 (mean of 12+3 is 7.5) and with B it went from 9 to 6, the mean of 9+6 is also 7.5. The post-hoc test is comparing these two means and saying there is no med*time difference between the groups rather than comparing the decrease with A versus B. Very confused now. Thanks!

[@VahidAryadoust]

To compare the decrease (rather than the mean of the variables), you will need to create new variables. Decrease from point 1 to 2 = point 1 - point 2. decrease from point 2 to 3 = point 2 - 3. etc. This will provide you with 4 new variables (decrease), which can be compared in the same way you have done.

[@philliptan8046]

This is super helpful, very grateful that u share this

[@gurkantarcin8536]

Thanks very much for this very informative video. Can I investigate "between subject effect" in Friedman test?

[@aengmuhidin9866]

Hello sir, what software are you using recording??

[@maramfaris6065]

Thank you, it was super helpful!

[@untonsured]

Hi, I am wondering, my p value in shapiro wilks is less than 0.05 in 4 of my conditions. What is the best thing to do here? I have 5 outliers also.

[@VahidAryadoust]

You could remove the outliers and try again. Also, set the alpha level at 0.01.

[@untonsured]

@@VahidAryadoust thanks. Would I have to justify using an alpha of .01?

[@untonsured]

@@VahidAryadoust I've got a good shapiro wilk alpha but now my levene's test is below .05 on one condition...

[@VahidAryadoust]

@@untonsured Some statistics books suggest that.

[@kennethlewis1516]

Could you share your spreadsheet of data, please?

[@tugbaaltintas2934]

Thank you, it was the best explanation.

[@alexisbarbry921]

Hi Sir, And thand you very much for this video. If there is no interaction between time and groups for example, I supposed that I can not go further on intra-group comparisons. Because for example, I have an interaction of time but no interaction of time*group. But when I go to the intra-group comparison, one group have statiscal differneces and the other have not... What do you think of that ?

[@VahidAryadoust]

Interaction is different from within-group or intra-group analysis. If there is no interaction, you still need to examine the within-group patterns.

[@benjaminhasselberg]

Also, where do you check for normality?

[@naveenmathieu]

query in stat analysis. i want to compare mean heart rate changes with baseline (T0) within the group.there are 10 continuous time points and the study group given at T2. (T0,T1,....T10.) like t1 vs t0, t2 vs t1, t3 vs t0.. what test should i apply ? paired t test is appropriate?.... or i should do Anova test

[@VahidAryadoust]

You should use repeated measures ANOVA.

[@VidhiS]

do i follow the same procedure if I have one within and two between factors?

[@VahidAryadoust]

Yes, you should.

[@besteuysal]

Hello, thank you for the useful video. I have a question. I have only two levels of my both within and between variables but Levene's test is significant so the assumption is not hold. What can I do to fix that? Thank you in advance.

[@VahidAryadoust]

hi, what is the p value of your Levene's test?

[@besteuysal]

@@VahidAryadoust I have two conditions and for one p=,002 and the other p=,011.

[@VahidAryadoust]

@@besteuysal Some scholars argue the alpha value should be set at 0.001 for the Levene's test.

[@benjaminhasselberg]

What if my between subjects factor has N = 57 in one group and N = 24 in the other?

[@bobd.28]

Hi sir, thank you for the clear explanation. I have to run a repeated measures ANOVA with age en gender, but in this case as covariates. What is the difference with implementing them as covariates instead of independent, between-subjects variables?

[@VahidAryadoust]

My understanding is that covariates are continuous (like age); so gender cannot be a covariate if it is binary.

[@susanshalani3569]

Hi Sir. Can I still interpret the within and between subject effects if I get a statistically significant for equality of covariance matrices? If not what should I do? My independent variable only has two levels.

[@VahidAryadoust]

Two schools of thought: 1: yes, you can, as ANOVA is a robust method; 2: no, you should use a non-parametric method in lieu.

[@susanshalani3569]

@@VahidAryadoust Thank you sir

[@Vaggos16]

I would like ti ask. What is the corelation coefficiente within subject and between when I have hierarchical data and how can I calculate It. Thanks in advance

[@VahidAryadoust]

If I understand your question correctly, you will need to use mixed effect models --> compute ICC.

[@Vaggos16]

@@VahidAryadoust I have a questionaire that each person has responded multiple times so my data is hierarchical and I want to calculate the corellation within person/subject and between. And this correlation isnt ICC

[@VahidAryadoust]

@@Vaggos16 I would use Pearson correlations.

[@fazeelzubair2787]

In my research I have two treatment 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?

[@VahidAryadoust]

I think a mixed-effect ANOVA (with between and within subject effects) is applicable in this case. treatment: between subject variable; time 90,2,4,8): within subject variable.

[@ibrahimouergui7318]

Hello, i wonder if you answer my question, i am testing the effect of two types of activities with the use of 3 ratio between time and rest and 2 rest intervals after the activities and the subsequent task. Which Anova should i use taking into consideration that i have a control condition in which none of all these variables were used, i though about repeated measure one way anova, howver if i would like to test the effect of the before-mentioned variables this will not be possible, any help ? thanks

[@VahidAryadoust]

It seems you will need to use repeated measures ANOVA, provided that your subjects were tested several times (i.e., a within-subject design).

[@ibrahimouergui7318]

@@VahidAryadoust thank you very much!! and if i include gender as a factor, this will be a mixed model anova ?

[@VahidAryadoust]

@@ibrahimouergui7318 YEs, a mixed model anova will do.

[@methodmatuba3712]

Hello i have one drug with two concentrations and i'm trying to their effectiveness at different time interval such as 0,2,4,6&8 how can i analyze this in spss

[@VahidAryadoust]

IT looks like a mixed ANOVA with drug type as the between subject variable, and time as the within subject variable).

[@methodmatuba3712]

@@VahidAryadoust thank

[@renelambino8668]

Hello sir, what version of SPSS are you using please??

[@VahidAryadoust]

This video was made using SPSS 25 (if I am not wrong). I am currently using SPSS27.

[@renelambino8668]

@@VahidAryadoust Thank you sir I appreciate it...

[@mouthofthesouth3903]

Hi, I’m not sure what test to conduct. I have 2 groups (experiments and control) and 10 samples in each group. All participants under go the same test before doing exercise and after doing exercise but the exercise from the control group is different to the experimental. What test is most appropriate

[@VahidAryadoust]

I think a repeated measures ANOVA with a between-subject variable (i.e., control vs experiment group) -- But please note that your sample size is pretty small. Perhaps use a non-parametric test.

[@courtneycollins1385]

Thank you, very helpful!

[@shanrdewi]

Would I use the same steps if I changed time 1 and time 2 into respectively: perceived effectivenes for leadership style A and leadership style B, moderated by gender?

[@VahidAryadoust]

The steps will be mostly the same.

[@shanrdewi]

@@VahidAryadoust Thankyou

[@tonjafras3652]

Hello I am doing a within- (8 levels) and between-subjects (5 levels) design repeated measures, but I get a warning that Box's Test of Equality of Covariance Matrices is not computed because there are fewer than two nonsingular cell covariance matrices. What seems to be the problem? What can I change? I still get the results but are they valid? Thanks!

[@VahidAryadoust]

I am not pretty sure, but it could be due to the sample size.

[@mishaalsharif5467]

Hi. Could you please help me with understanding the following? 1. What does it mean if there is no interaction between gender*time (within subjects)? Does it mean that amongst the females there are no differences across the time points? 2. If there was a between subjects effect of gender, does that mean there is a significant difference of time between males and females? Thanks!

[@VahidAryadoust]

For both the questions, your interpretation is correct..

[@minahabibi1008]

thank you so much. I have a question, what if we have some missing value in some levels?

[@VahidAryadoust]

The missing values will be removed listwise by SPSS.

[@syaidatulafiqah3206]

Sir, I have a question. How about if I have more than 2 variables and want I to differentiate these variables? Can I use this test?

[@VahidAryadoust]

If you have measured the variables at least twice over time, yes you can. Otherwise, you should apply other types of analysis.

[@syaidatulafiqah3206]

@@VahidAryadoust oh I see, okay thank you sir

[@fakuridesne]

Tq very much. Very2 helpful

[@milanagorelik4624]

Hello, thank you for the video! I have a question - I have 2 groups (experiment and control). Each group is comprised of males and females which are measured on the dependent variable 3 times. My independent variables are group, gender and time. Will repeated measures ANOVA work for this study design? Any help will be greatly appreciated!

[@VahidAryadoust]

Try a mixed ANOVA, which combines both repeated measures ANOVA and factorial ANOVA. Your gender and group variables are between group, and time is within group.

[@milanagorelik4624]

@@VahidAryadoust Thank you for your prompt answer! 🙏🏻

[@جماليحي-ث8ق]

Inquiry: If I had one group of males and females, and I applied treatment to them and took three measurements If the conditions for a t-test are present, use a two-sample t-test to test the treatment difference between males and females in each measurement Secondly, it is possible to use a repeated-measures test to compare the effect of treatment on males in repeated measurements I repeat this again for females

[@neethups518]

Can u provide a final table format?

[@VahidAryadoust]

Sorry, table format for purpose?

[@neethups518]

@@VahidAryadoust For including in the research paper

[@VahidAryadoust]

@@neethups518 this video will give you some idea how to report the results: kzbin.info/www/bejne/bmWcfY1oaat1e6M

[@AJP0987654321]

Thank you

[@chacmool2581]

Excel? 😂😂😂