This is an EXCELLENT explanation. Thank you very much! I am ALWAYS looking for explanations like this that EXPLAIN the big picture, BEFORE getting into the weeds of, "here's how to do all these computations...". Excellent. Thank you!
Just how beautifully Prof Ray can introduce the concepts underlying what most of us perceive as "complex statistical mumbo jumbo", is a lesson in its own right for teachers in particular. Highly recommended!!
Thanks Brother! Wow, that was a mouthful! However, this short video has given me the answer I've been looking for. Thanks for explaining such a complex issue in the most simple terms.
Hi. Thanks for the explanation! Finally someone who could explain to me in simple English about the concept of Component Analysis without the statistical jargon!
This was amazing! Best part was the explanation of the difference between common factor analysis and principle component analysis. Also, eigenvalues. Thank you
Ellen, Yes - just rerun your component analysis and select regression factor score in the SPSS Factor procedure (accessed via the 'Scores' button) and indicate that you want a factor score saved. This will create a single score on your one component that comprises contributions from each variable - the highest loading variables will contribute most to the score. I would only recommend this procedure for a single component solution, however. Cheers, Ray
hi, mr ray cooksey thank you very mush i study cpa in frensh language (+ all the mathematics cours :'( ) but i didnt understand mush thing but with your video its really helpfull even with my bad english i get the principals things about CPA and FA. thanks again
Thank you so much and keep going on with Fantastic videos like this ....the only thing i think you missed is to write the head lines of the main subject in an outline fashion so the interpretation would be much easier but thanks any way Great job
If you mean what would yo do to extract just one factor, I would use SPSS factor analysis with principal components extraction and force the procedure to produce a one-component solution. no need for rotation. The loading matrix would tell you how strongly each variable is aligned with that single component. Again, if you are only interested in one component, then you could use SPSS to produce a regression factor score (a type of z-score) for each respondent in your data file. Cheers, Ray
Principal components analysis is an exploratory factor analytic procedure, where the factor/component structure of the variables is not hypothesised in advance. Confirmatory factor analysis requires the researcher to hypothesise, in advance, what he/she thinks the factor structure of the variables is and provides that structure as input into the analysis. Confirmatory factor analysis then provides statistics to assess the goodness-of-fit of that hypothesised structure to the data. Ray Cooksey
Hi Ray, thanks for the prompt reply!! I am looking at creating an aggregate score of subjective well-being using the PANAS and SWLS scales. I used the method you suggested and it worked great. Would you obtain a regression factor score as well or is this an alternative method? Thanks again! :)
Thank you very much Pr Cooksey. Can we use PCA or FA to weight indicators of a variables. My variables have each one a set of indicators and I would like to know how to weight those indicators. PS: the indicators have different units (currency, kg, numbers....)
Great lecture! So I have a question. I've conducted a likert-scale survey (the MSI/ Maslach Burnout Inventory survey) and obtained 501 responses. MSI is supposed to meassure burnout in 3 dimensions (emotional exhaustion, depersonalization and personal achievement). After running pca with R I get a p value of 2.59e-25 which seems to be very small, compared to other studies. I'm I missing something? I don't seem to understand how to interpret this result. Thanks a lot for the help!
This is a fantastic explanation!! THANK YOU! :) Can you please explain to me if I am planning on specifying for a one factor extraction what is the process I should be using? Cheers.
Hi there. I have one dichotomous dependent variable and 30 independent variables. Originally, I planned to conducted binary logistic (BLR) regression to see the effects of independent variables on the dependent variable. Later I encountered an issue that for BLR we can not use more than 10 independent variables. So, I planned to conduct factor analysis first to reduce 30 independent variables into few factors. Then, I conducted BLR considering those reduced factors and dependent variable. Did I do it right?
One thought on how this could be made even better. Add some examples. Use the EXACT SAME content / verbiage as in this video, but show some examples of some past studies demonstrating the concepts in real life. (Instead of just looking at your face and hands the whole time....) :-)
This is an EXCELLENT explanation. Thank you very much! I am ALWAYS looking for explanations like this that EXPLAIN the big picture, BEFORE getting into the weeds of, "here's how to do all these computations...". Excellent. Thank you!
I am a Psych student at UNE. Ray, I wish you were the stats teacher for us. You have a gift. I think you just saved my degree!
Excellent clarity, without using a single slide !
Just how beautifully Prof Ray can introduce the concepts underlying what most of us perceive as "complex statistical mumbo jumbo", is a lesson in its own right for teachers in particular. Highly recommended!!
Thanks for your selfless dedication to sharing your knowledge Dr. Cooksey. This was a terrific lecture!
Thanks Brother!
Wow, that was a mouthful! However, this short video has given me the answer I've been looking for. Thanks for explaining such a complex issue in the most simple terms.
Thank you very much, sir ! Great explanation.
Thanks Ray. By explaining such a complicated concept with simple language you showed you are a real pedagogue in statistics! :)
Hi. Thanks for the explanation! Finally someone who could explain to me in simple English about the concept of Component Analysis without the statistical jargon!
This was amazing! Best part was the explanation of the difference between common factor analysis and principle component analysis. Also, eigenvalues. Thank you
I read a journal article that used factor analysis but was a bit confused about what was going on. This was very helpful. Thank you for great content!
Nice, engaging explanation. It really helped me tie together the concepts talked about in my lectures
You did a great job simplifying the logic of such a complex statistical analysis; I hope you keep posting similar videos!
Excellent explanation. Bravo!
Ellen, Yes - just rerun your component analysis and select regression factor score in the SPSS Factor procedure (accessed via the 'Scores' button) and indicate that you want a factor score saved. This will create a single score on your one component that comprises contributions from each variable - the highest loading variables will contribute most to the score. I would only recommend this procedure for a single component solution, however. Cheers, Ray
Thank you! This was so clear and you made it so much easier to understand. :)
You are a great teacher... thank you very much
hi, mr ray cooksey
thank you very mush
i study cpa in frensh language (+ all the mathematics cours :'( ) but i didnt understand mush thing
but with your video its really helpfull even with my bad english i get the principals things about CPA and FA. thanks again
Oustanding explanation of core concepts.
Just sums it all up, stripped of th clutter.
That's your Eigenvalue right there
That's very great explanation. Thank you Professor.
thanks a lot sir !! all doubts cleared in 15 mins.
Super helpful. Thank you.
my God what an excellent explanation! thank you!
Thank you so much and keep going on with Fantastic videos like this ....the only thing i think you missed is to write the head lines of the main subject in an outline fashion so the interpretation would be much easier but thanks any way Great job
Fantastic! Where were you when I was taking Stats?!! I'm looking forward to watching your other videos!
thank you for making these videos, really helpful!
Awesome!! Great explanation:)
Thank you so much!!! Nice explanation. I really needed this
If you mean what would yo do to extract just one factor, I would use SPSS factor analysis with principal components extraction and force the procedure to produce a one-component solution. no need for rotation. The loading matrix would tell you how strongly each variable is aligned with that single component. Again, if you are only interested in one component, then you could use SPSS to produce a regression factor score (a type of z-score) for each respondent in your data file. Cheers, Ray
Is Principal Component Analysis is the came as Confirmatory Factor Analysis?
PS: Thanks for the video. It explains everything very clearly.
Excellent explanation! Thank you so much and please keep up the great videos! Could you please do one on G-theory?
Principal components analysis is an exploratory factor analytic procedure, where the factor/component structure of the variables is not hypothesised in advance. Confirmatory factor analysis requires the researcher to hypothesise, in advance, what he/she thinks the factor structure of the variables is and provides that structure as input into the analysis. Confirmatory factor analysis then provides statistics to assess the goodness-of-fit of that hypothesised structure to the data. Ray Cooksey
excellent explanation!!!
Hi Ray, thanks for the prompt reply!! I am looking at creating an aggregate score of subjective well-being using the PANAS and SWLS scales. I used the method you suggested and it worked great. Would you obtain a regression factor score as well or is this an alternative method? Thanks again! :)
Thank you very much Pr Cooksey. Can we use PCA or FA to weight indicators of a variables. My variables have each one a set of indicators and I would like to know how to weight those indicators.
PS: the indicators have different units (currency, kg, numbers....)
Great lecture! So I have a question. I've conducted a likert-scale survey (the MSI/ Maslach Burnout Inventory survey) and obtained 501 responses. MSI is supposed to meassure burnout in 3 dimensions (emotional exhaustion, depersonalization and personal achievement). After running pca with R I get a p value of 2.59e-25 which seems to be very small, compared to other studies. I'm I missing something? I don't seem to understand how to interpret this result. Thanks a lot for the help!
This is a fantastic explanation!! THANK YOU! :)
Can you please explain to me if I am planning on specifying for a one factor extraction what is the process I should be using? Cheers.
Thank you so much.
Just perfect
Thank you Sir!!
Hi there. I have one dichotomous dependent variable and 30 independent variables. Originally, I planned to conducted binary logistic (BLR) regression to see the effects of independent variables on the dependent variable. Later I encountered an issue that for BLR we can not use more than 10 independent variables. So, I planned to conduct factor analysis first to reduce 30 independent variables into few factors. Then, I conducted BLR considering those reduced factors and dependent variable. Did I do it right?
thank you for the explanation
Its excellent.
What is meant by "variance" in this case?
Thank You!
Sample variance.
Any hints for us using "R" and not "SPSS"...
One thought on how this could be made even better. Add some examples. Use the EXACT SAME content / verbiage as in this video, but show some examples of some past studies demonstrating the concepts in real life. (Instead of just looking at your face and hands the whole time....)
:-)
11:00 Rotation
THIS is how a factor analysis help video should be done! Straight to the point and oh SO helpful! Thank you so much!!!!!