Really great and useful video - thanks for sharing your knowledge! For my model I’m using multiple dependent variables that have different units (reaction time, rate, total scores,…). Is that a problem?
How to do Manova for 1 independent and more than 3 depndent variables ( Panel data set)... Kindly reply ASAP. Thanks Please provide emprical model for above case as well. Thanks a lot
Thanks a lot! How is it that at 4:48 you can call the TRT and REP column of your dataset without specifying the data it is in? I would expect the line of code to read "treat_factor
Hi, I found this really useful and could follow up to the end bit. I only have two dependent variables so what do I do when you're doing the matrix bit at the end because I only have two. I want to get that interaction table between the dependent variables like you have. Thank you!!!
Do you have an independent variable as well? You need that to perform your hypothesis tests as well. Otherwise, can you provide me a timestamp for additional context?
@@SpencerPaoHere Thanks so much for getting back to me!! I have two independent (gender: female, male and non-binary and body ideal: thin ideal and muscular ideal) variables in my study and two dependent variables (eating disorder awareness and help seeking). I want to look at the interaction between the independent variables on these dependent variables. But most importantly what I am doing the MANOVA for is the interaction between my dependent variables, eating disorder awareness and help seeking. I was following your video up until 8 minutes 48 seconds. I wondered if you any tips or r code you could give me to help me with the results I need as I am struggling. Thank you!!!
@@jessstace9895 Sounds like you are attempting to find the some form of interaction between the independent and dependent variables? Have you attempted to run summary.aov(model)? 8:04 If you want to find multiplicative relationships or additional additive relationships, you will have to edit your model formula (y~x) where the x variables can be Feature_1*Feature_2 + Feature_3 ... to then find additional relationships combined to identify a dependent variable causation if any. You can refer to line 32 in the video (around 5:28) for additional details. I hope this helps!
I'm glad you liked it! There could be a few issues related to that. Your datatypes are not consistent. Try to convert everything to a numeric using as.numeric() - so datatypes are consistent. Also, it looks like your rows might not add up? Else, send me some reproducible code and I can help from there.
Hi! That can be a multitude of reasons. Try This: par(mar=c(5,4,4,2)) Or, you can increase the resolution on your screen-- also expand your plotting screen for increased pixels (and try running the plots again) Or, you can navigate to Plots > Remove Plots on the Rstudio GUI. That should reset the default par parameters.
@@SpencerPaoHere Hi, it's me again. I am confused with what you have said in 12:53. What do you mean by "their specific observations in our treatment are NOT VALUABLE to our overall model"? Would that mean that these specific observations do not statistically differ in terms of the dependent variables? (I'm sorry for the capitalized phrase, I just want to emphasize on that).
@@pauljohncapote7541 So, when you are testing some hypothesis (matrix(c(x,y,z...)), that allows you to test more than 1 restriction on your coefficients. So when it comes to the not valuable part -> You can use the hypothesis testing to test whether or not some coefficient (c(x,y,z...) has a significant effect on your model.
Thanks! But at 8:13 I think you make a mistake by stating treatment factor has no effect when its P value was lower than the representation factor?
Nice catch! You are definitely correct to point that out. I misread the statistic and thought that the P value was > 0.05
Thank you for doing the video! It’s really helpful.
Great video! I could carry my analysis from start to end without any issue! Thank you so very much for the help!
Really appreciate the video! Super helpful in interpreting results AND how to apply the code!
This is very helpful. Thank you!
This was super helpful and easy to follow! Thank you so much for helping me get through my thesis!
Awesome cheers from Brazil
Muito bom! Excelente explicação. Obrigado.
Thanks!
This was very useful! thanks!
Thank you a lot for the video! It really helped :)
Really great and useful video - thanks for sharing your knowledge! For my model I’m using multiple dependent variables that have different units (reaction time, rate, total scores,…). Is that a problem?
Nope!
How to do Manova for 1 independent and more than 3 depndent variables ( Panel data set)... Kindly reply ASAP. Thanks Please provide emprical model for above case as well. Thanks a lot
Thanks a lot! How is it that at 4:48 you can call the TRT and REP column of your dataset without specifying the data it is in?
I would expect the line of code to read "treat_factor
The "attach(data_set_name)" allows you to call the dataset variables without calling the dataset$feature_name.
Hi, I found this really useful and could follow up to the end bit. I only have two dependent variables so what do I do when you're doing the matrix bit at the end because I only have two. I want to get that interaction table between the dependent variables like you have. Thank you!!!
Do you have an independent variable as well? You need that to perform your hypothesis tests as well. Otherwise, can you provide me a timestamp for additional context?
@@SpencerPaoHere Thanks so much for getting back to me!! I have two independent (gender: female, male and non-binary and body ideal: thin ideal and muscular ideal) variables in my study and two dependent variables (eating disorder awareness and help seeking). I want to look at the interaction between the independent variables on these dependent variables. But most importantly what I am doing the MANOVA for is the interaction between my dependent variables, eating disorder awareness and help seeking. I was following your video up until 8 minutes 48 seconds. I wondered if you any tips or r code you could give me to help me with the results I need as I am struggling. Thank you!!!
@@jessstace9895 Sounds like you are attempting to find the some form of interaction between the independent and dependent variables? Have you attempted to run summary.aov(model)? 8:04
If you want to find multiplicative relationships or additional additive relationships, you will have to edit your model formula (y~x) where the x variables can be Feature_1*Feature_2 + Feature_3 ... to then find additional relationships combined to identify a dependent variable causation if any. You can refer to line 32 in the video (around 5:28) for additional details. I hope this helps!
thank you, is it possible;le to share the code
Hi, Thanks for the great tutorial and I am now well on my way to performing my own MANOVA. I've stalled at the moment as I can't run:
mod
I'm glad you liked it! There could be a few issues related to that.
Your datatypes are not consistent. Try to convert everything to a numeric using as.numeric() - so datatypes are consistent.
Also, it looks like your rows might not add up?
Else, send me some reproducible code and I can help from there.
@@SpencerPaoHere That's helped me to progress. Many thanks
I have problem with Boxplots. Every time I run the codes, it says "Error in plot.new(): figure margins too large", could you help me, please?
Hi! That can be a multitude of reasons.
Try This: par(mar=c(5,4,4,2))
Or, you can increase the resolution on your screen-- also expand your plotting screen for increased pixels (and try running the plots again)
Or, you can navigate to Plots > Remove Plots on the Rstudio GUI. That should reset the default par parameters.
@@SpencerPaoHere I will do that, thank you Sire.
@@SpencerPaoHere Hi, it's me again. I am confused with what you have said in 12:53. What do you mean by "their specific observations in our treatment are NOT VALUABLE to our overall model"? Would that mean that these specific observations do not statistically differ in terms of the dependent variables? (I'm sorry for the capitalized phrase, I just want to emphasize on that).
@@pauljohncapote7541 So, when you are testing some hypothesis (matrix(c(x,y,z...)), that allows you to test more than 1 restriction on your coefficients. So when it comes to the not valuable part -> You can use the hypothesis testing to test whether or not some coefficient (c(x,y,z...) has a significant effect on your model.