I really appreciate for your fluent and clear-cut presentation! I will be thankful if you help me find a way to do the analysis of my research. The three treatment groups of writing skill development are scored analytically in their pre- and post-tests leading to 5 scores for 5 different writing traits. How could I analyze and interpret the effect of these three models (IV) on any of these five traits (DV)?
If you want to treat the 5 traits separately (as stand-alone outcomes) you could perform the analysis as a MANOVA in order to analyze the effect of the IVs on all 5 traits simultaneously and then separately. This is especially effective if the 5 traits are related to each other.
One question: To observe statistical significance between the 3 different time points (time effect) you mentioned to observe multivariate test at the "time" row and Wilks Lambda test and observe at significance column "Sig" =0.000.This means time is significant. Is this time effect significance valid for both groups? what happens if in one group the change during time points is significant but in the other group it is not significant?. I would like to know the time effect for each group but also If the effect of the 2 different types of treatment between groups is statistically significant
+Juan Carlos The Time effect is for all subjects regardless of group. If there is a differential change in groups over time that is significant, then the Interaction effect will be significant.
Hi thanks for the video, is really helpful. I'd like to know if there is any video explaining the additional tests (contrasts) in the case we found statistical significant for the interaction in the multivariate tests?
Thank you for this helpful video. What if you have this situation: You want to compare outcomes of two groups(control & intervention) across three time periods(baseline, intervention, and post-baseline). Baseline data were collected everyday for 6 days, the intervention data were collected for 30 days, and post-baseline data were collected for 12 days. How to organize and analyze the data??
I suppose the most straightforward approach would be to create a mean score of the outcome for each time period (i.e. mean of 6 baseline days, mean of 30 intervention days, etc). I think it would be very complicated to treat all 48 days that data was collected as separate time points, so you will probably have to aggregate your outcome data in some way.
Thank you for this very helpful and eloquent video. However, I still have some questions. 1. How can I measure the difference in difference (mean with 95%CI) between 2 groups at 2 different time intervals? Using paired T test I am able to determine the mean (with 95%CI) difference in each group (at baseline and after treatment), and with this analysis (split-plot ANOVA) I can say that the effect of the treatment is not different between this two groups (from the Test of Between Subjects Effects). But it doesn't show this difference in difference. 2. What should I do if the Levene's test of Equality of Error Variances is significant? Thank you!
1. Are you referring to the actual difference in the means of the two groups? I'm not clear on what you are asking for. 2. You can still run the analysis but it becomes a limitation that should be acknowledged in your results section.
This was very helpful, thank you. I added covariates to my analysis. In the "Test of Between Subjects Effects," what does it mean when the covariate is
Hey there. I'm measuring how the abundance at both Patch A and Patch B changes over time (20 second increments (0-300s)). I have 10 trials worth of data. I'm a little lost on how to structure this on the spreadsheet. Any help would be great, thanks!
Thanks a lot for the video! Unfortunately I don't really get the different between time * group (multivariate tests) and the group (in the between subject)'s sig. In the multivariate tests table, as you have mentioned, the time * group is not significant, could you please elaborate me what exactly does that mean? What if it was significant?
The between subjects significance measures the effect of the group assignment on the outcome by itself, regardless of time. The time*group interaction looks for a unique combination of group and time point that explains more change/difference than other combinations. If the interaction is significant that makes it very difficult to determine the effect of group or time on the outcome independently.
Thank you very much for the information. If that happens, I mean the significance in both time*group and group, what should one do? Let's say by having 2 group and 3 repeated measures, if both time*group and group were significant, Is there anything that I can do to determine the effect of the group independently? And am I allowed to claim the significant effect of the treatment (group 2)?
HichkasNist Yes, if there is a significant interaction you can perform a post-hoc test called Contrasts. That will allow you to tease out which combination of time and group is significant. However, it precludes you from definitively making conclusions about the effect of the treatment by itself.
You will have to analyze each pair of groups to determine statistical significance. The post-hoc test will produce a table that will allow you to interpret the p value for each pair-wise comparison.
Hello, great video! How can I make a single diagramm in SPSS with 5 personality factors with 2 levels of independent variables? It is an ANOVA repeated measurement.
Thank you very much for the nice tutorial but I have a questions: 1) how to make an error bars to appear now on this line plots? (the option in chart editor `show the error bars` cant be clicked ) 2) if I have 6 groups of treatment which post hoc test could be used if you might know?
Hi i have a question that what if I have 3 groups for 3 products? and I have 39 timepoints, i want to know which product has most significant effect, how can I analyse them? Thank you !
Is each group receiving only one of the products? or is each group receiving different combinations of products? If each group is receiving only one product then you will have 3 groups for one independent variable. If each group is receiving different combinations of products then you will have as many groups as you do combinations of products.
Hi, thank you for the great video. It is so clear and useful and it is exactly like me study. I have one question though, in the estimated marginal means is there a way to compare the mean at each time factor for each group alone? Thanks again :)
Hi is me again! What can I do if the Levine´s test gave me statistical differences (meaning the variance are not homogenic) Im comparing 7 treatments (groups) and 4 repeated measures! Help :) Thank you
Do the same steps apply if I have 2 treatment groups (temperature and snow depth) both with 2 levels (enhanced and control), measure at 2 time points? I would just make 2 columns, one for each group, correct?
can you still use a factorial repeated measures Anova if there are only two depression time points instead of 3? If not which ANOVA could you use? thanks
If my dependent variable would be ordinal (with 5 different levels), can i possible conduct a mixed between-within subjects ANOVA? I know it violates the assumption of normality but i have an exam question that is formulated that way :/
Mate, I've been struggling. I got 21 samples (so 1 to 21) and each sample has 7 different concentrations (one of this concentrations is the control). I tried to put the different concentrations as variables, but it didn't work. Then, I also put the samples as variables (so 21 variables) and the concentrations as subjects (7 lines). Nothing works... any tips? cheers. Great videos btw
+Killian Maciel It sounds as though you should have one column coded for your 7 different concentrations and then another column for your samples. The second column will have 21 sample outcomes (rows) for concentration #1, then 21 more rows for concentration #2 and so on.
hello, can you please show me how to perform 2x2x8 Mixed ANOVA in SPSS between subject factor 2 levels 1)CAI 2) Control, within subject factor 2 levels 1) Injured 2) Uninjured , with 8 levels of directions (AL,A,AM,M,PM,P,PL,L) I am Sports Physio. & i am doing study on tennis players with ankle instability for my Thesis ( masters). I have done the experiment, I have collected the data, but i am facing huge difficulty in feeding it into SPSS and analyzing it. Can you please help me with it. So that i can submit
+Zubin Bhavsar Were all subjects exposed to all 8 of your directions? Each subject row will have to have a level of all of the independent variables associated with it. For example subject 1 is a control, is injured and has direction AL. So at minimum your data file will have 4 columns, one for between subject factor, one for within subject factor, one for direction and one for the outcome.
Thank you for this useful video. I have a question though. I'm doing a study with 4 groups of mice (5 each) where the effect is measured across 5 different time points for each of the mouse and for each of the group. one of the groups is my control. So, this rmANOVA tells me which groups are significant compared to control (with dunnett) and with bonferroni, i can compare each groups with other or each time point with other. BUT, what I want to know, which of the test groups significantly differ from the control group AT THE SAME time point...that is, i want to compare the means of the test groups with the control group at the same time point..is that possible?
***** Yes, I already did that. But, I encountered some journal, where they did rmANOVA for the situation i described earlier...so, is it possible to do a repeated measures ANOVA for that situation? Im guessing the answer is no...but, how did they do it then, i cant seem to find the answer to that anywhere.
Rakibul Islam If you are only interested in determining group differences at one time point, then I don't see the necessity of performing a RM ANOVA.. If you want to compare differences between your 5 time points regardless of group membership, then RM ANOVA would be appropriate.
I performed a test like this however I needed to transform my data from my descriptive box (which I then back-transformed). How would this affect the numbers on the axis of my graph to view the data?
Hi I was just wondering if it's possible to run a three-way repeated measures ANOVA? I have 2 Treatment groups each subjected to 7 levels of another categorical variable which is repeated at 4 different levels of another categorical variable which is also repeated. There is one continuous dependent variable from all of this. Would I have to split up the treatment groups into two 2-way repeated ANOVAs? or is there a way to test all three factors at once? Thanks
+Christian1722 Yes, you can. If your 7 and 4 level categorical variables are repeated in the same manner for all subjects then those could be treated as within-subject factors and treatment groups would be your between subjects factor.
Great work! Been very helpful. Couldn't be more grateful
Thank you for your outstanding lecture.
This content was extremely helpful to my clinical study.
Thanks a lot. Hopefully you can do an example with 3 grps or more. But this was clear and precise. Good Job.
Thank you! This tutorial helped me for my master thesis!
Great presentation,
Very helpful.
Thank you!
I really appreciate for your fluent and clear-cut presentation! I will be thankful if you help me find a way to do the analysis of my research. The three treatment groups of writing skill development are scored analytically in their pre- and post-tests leading to 5 scores for 5 different writing traits. How could I analyze and interpret the effect of these three models (IV) on any of these five traits (DV)?
If you want to treat the 5 traits separately (as stand-alone outcomes) you could perform the analysis as a MANOVA in order to analyze the effect of the IVs on all 5 traits simultaneously and then separately. This is especially effective if the 5 traits are related to each other.
Wow! great video sir! thank you very much!
Thanks for this. Really useful and well explained! :)
Really good video! Thanks for sharing it :)
I am stuck - I don't understand why my Mauchly's test of spherecity is a . value and has no numbers at present at all!?
One question: To observe statistical significance between the 3 different time points (time effect) you mentioned to observe multivariate test at the "time" row and Wilks Lambda test and observe at significance column "Sig" =0.000.This means time is significant. Is this time effect significance valid for both groups? what happens if in one group the change during time points is significant but in the other group it is not significant?. I would like to know the time effect for each group but also If the effect of the 2 different types of treatment between groups is statistically significant
+Juan Carlos The Time effect is for all subjects regardless of group. If there is a differential change in groups over time that is significant, then the Interaction effect will be significant.
Hi thanks for the video, is really helpful. I'd like to know if there is any video explaining the additional tests (contrasts) in the case we found statistical significant for the interaction in the multivariate tests?
Alex Santi Yes. ua-cam.com/video/v_NHBzlnDS8/v-deo.html
Thank you for this helpful video. What if you have this situation: You want to compare outcomes of two groups(control & intervention) across three time periods(baseline, intervention, and post-baseline). Baseline data were collected everyday for 6 days, the intervention data were collected for 30 days, and post-baseline data were collected for 12 days. How to organize and analyze the data??
I suppose the most straightforward approach would be to create a mean score of the outcome for each time period (i.e. mean of 6 baseline days, mean of 30 intervention days, etc). I think it would be very complicated to treat all 48 days that data was collected as separate time points, so you will probably have to aggregate your outcome data in some way.
*****
Thank you! That is exactly what I was considering about!
Thank you for this very helpful and eloquent video. However, I still have some questions.
1. How can I measure the difference in difference (mean with 95%CI) between 2 groups at 2 different time intervals?
Using paired T test I am able to determine the mean (with 95%CI) difference in each group (at baseline and after treatment), and with this analysis (split-plot ANOVA) I can say that the effect of the treatment is not different between this two groups (from the Test of Between Subjects Effects). But it doesn't show this difference in difference.
2. What should I do if the Levene's test of Equality of Error Variances is significant?
Thank you!
1. Are you referring to the actual difference in the means of the two groups? I'm not clear on what you are asking for.
2. You can still run the analysis but it becomes a limitation that should be acknowledged in your results section.
This was very helpful, thank you. I added covariates to my analysis. In the "Test of Between Subjects Effects," what does it mean when the covariate is
+Carlos Almeida Then the covariate has a significant effect on the outcome.
thank you! Very helpful :)
Hey there. I'm measuring how the abundance at both Patch A and Patch B changes over time (20 second increments (0-300s)). I have 10 trials worth of data. I'm a little lost on how to structure this on the spreadsheet. Any help would be great, thanks!
+Rhys Diplock Much like I have it in the example; one column denoting patch a and b and then 10 columns denoting your trials.
Thanks a lot for the video!
Unfortunately I don't really get the different between time * group (multivariate tests) and the group (in the between subject)'s sig.
In the multivariate tests table, as you have mentioned, the time * group is not significant, could you please elaborate me what exactly does that mean? What if it was significant?
The between subjects significance measures the effect of the group assignment on the outcome by itself, regardless of time. The time*group interaction looks for a unique combination of group and time point that explains more change/difference than other combinations. If the interaction is significant that makes it very difficult to determine the effect of group or time on the outcome independently.
Thank you very much for the information.
If that happens, I mean the significance in both time*group and group, what should one do?
Let's say by having 2 group and 3 repeated measures, if both time*group and group were significant, Is there anything that I can do to determine the effect of the group independently? And am I allowed to claim the significant effect of the treatment (group 2)?
HichkasNist Yes, if there is a significant interaction you can perform a post-hoc test called Contrasts. That will allow you to tease out which combination of time and group is significant. However, it precludes you from definitively making conclusions about the effect of the treatment by itself.
***** Thank you for the detailed information!
Thank you for this video!
Could you please give me a few details with respect to the post hoc test when there are three groups in the study?
You will have to analyze each pair of groups to determine statistical significance. The post-hoc test will produce a table that will allow you to interpret the p value for each pair-wise comparison.
Hello, great video! How can I make a single diagramm in SPSS with 5 personality factors with 2 levels of independent variables? It is an ANOVA repeated measurement.
+operationcenterbravo Not sure what you mean by diagram?
Thank you very much for the nice tutorial but I have a questions:
1) how to make an error bars to appear now on this line plots? (the option in chart editor `show the error bars` cant be clicked )
2) if I have 6 groups of treatment which post hoc test could be used if you might know?
1. I'm not sure why the error bar option is not showing up for you.
2. I would suggest the Bonferroni post-hoc.
Hi i have a question that what if I have 3 groups for 3 products? and I have 39 timepoints, i want to know which product has most significant effect, how can I analyse them? Thank you !
Is each group receiving only one of the products? or is each group receiving different combinations of products?
If each group is receiving only one product then you will have 3 groups for one independent variable.
If each group is receiving different combinations of products then you will have as many groups as you do combinations of products.
Hi, thank you for the great video. It is so clear and useful and it is exactly like me study. I have one question though, in the estimated marginal means is there a way to compare the mean at each time factor for each group alone? Thanks again :)
+nar2086 You can ask for the means for the Interaction for each group and time point and means will be provided as you describe.
+TheRMUoHP Biostatistics Resource Channel Thank you for the explanation.
Hi is me again! What can I do if the Levine´s test gave me statistical differences (meaning the variance are not homogenic) Im comparing 7 treatments (groups) and 4 repeated measures! Help :) Thank you
You would decrease the p value (make it more stringent) that you are using to determine statistical significance. I would suggest using p
Do the same steps apply if I have 2 treatment groups (temperature and snow depth) both with 2 levels (enhanced and control), measure at 2 time points? I would just make 2 columns, one for each group, correct?
+sean arruda Yes, that is correct.
+TheRMUoHP Biostatistics Resource Channel
Where would you include the extra independent variable when you're defining the variables?
can you still use a factorial repeated measures Anova if there are only two depression time points instead of 3? If not which ANOVA could you use? thanks
+richard saliba Yes, you could. As long as each independent variable has at least two levels, this technique can be used.
If my dependent variable would be ordinal (with 5 different levels), can i possible conduct a mixed between-within subjects ANOVA? I know it violates the assumption of normality but i have an exam question that is formulated that way :/
+Fredrik Segerberg You can technically use that analysis but it would have the limitation of non-normal data. I would not recommend it.
Mate, I've been struggling. I got 21 samples (so 1 to 21) and each sample has 7 different concentrations (one of this concentrations is the control). I tried to put the different concentrations as variables, but it didn't work. Then, I also put the samples as variables (so 21 variables) and the concentrations as subjects (7 lines). Nothing works... any tips? cheers. Great videos btw
+Killian Maciel It sounds as though you should have one column coded for your 7 different concentrations and then another column for your samples. The second column will have 21 sample outcomes (rows) for concentration #1, then 21 more rows for concentration #2 and so on.
this is amazing. but do you have a sample data set for spss for this problem?
+Nichelle Nudalo I do, send me your email address via message and I will send it to you.
hello, can you please show me how to perform 2x2x8 Mixed ANOVA in SPSS
between subject factor 2 levels 1)CAI 2) Control, within subject factor 2 levels 1) Injured 2) Uninjured , with 8 levels of directions (AL,A,AM,M,PM,P,PL,L)
I am Sports Physio. & i am doing study on tennis players with ankle instability for my Thesis ( masters). I have done the experiment, I have collected the data, but i am facing huge difficulty in feeding it into SPSS and analyzing it. Can you please help me with it. So that i can submit
+Zubin Bhavsar Were all subjects exposed to all 8 of your directions? Each subject row will have to have a level of all of the independent variables associated with it. For example subject 1 is a control, is injured and has direction AL. So at minimum your data file will have 4 columns, one for between subject factor, one for within subject factor, one for direction and one for the outcome.
Thank you for this useful video. I have a question though. I'm doing a study with 4 groups of mice (5 each) where the effect is measured across 5 different time points for each of the mouse and for each of the group. one of the groups is my control. So, this rmANOVA tells me which groups are significant compared to control (with dunnett) and with bonferroni, i can compare each groups with other or each time point with other. BUT, what I want to know, which of the test groups significantly differ from the control group AT THE SAME time point...that is, i want to compare the means of the test groups with the control group at the same time point..is that possible?
Yes, you could do a one-way ANOVA using your 4 groups as your independent variable/factor and your one time point as your dependent variable/outcome.
***** Yes, I already did that. But, I encountered some journal, where they did rmANOVA for the situation i described earlier...so, is it possible to do a repeated measures ANOVA for that situation? Im guessing the answer is no...but, how did they do it then, i cant seem to find the answer to that anywhere.
Rakibul Islam If you are only interested in determining group differences at one time point, then I don't see the necessity of performing a RM ANOVA.. If you want to compare differences between your 5 time points regardless of group membership, then RM ANOVA would be appropriate.
I performed a test like this however I needed to transform my data from my descriptive box (which I then back-transformed). How would this affect the numbers on the axis of my graph to view the data?
As long as the data was back-transformed, you should be able to graph it.
***** Thank you :)
Hi I was just wondering if it's possible to run a three-way repeated measures ANOVA? I have 2 Treatment groups each subjected to 7 levels of another categorical variable which is repeated at 4 different levels of another categorical variable which is also repeated. There is one continuous dependent variable from all of this. Would I have to split up the treatment groups into two 2-way repeated ANOVAs? or is there a way to test all three factors at once? Thanks
+Christian1722 Yes, you can. If your 7 and 4 level categorical variables are repeated in the same manner for all subjects then those could be treated as within-subject factors and treatment groups would be your between subjects factor.