If anyone is searching all over RUclips for a video explaining SEM in a simplified yet concised manner, then this one's it 👌🏼👌🏼👌🏼👍🏼..worth 40 mins of your life.
This lecture series is amazing. I had to recently take a module on my PhD on Quantitative Methods, and as a first time learner of SEM, I find this video really helpful. Thank you very much Professor Sturgis.
The only productive critic that I would make is that while having a real person presenting the topic is perhaps more entertaining for the viewer, it adds a bit of extraneous load on them. As per Mayer's recommandations on multimedia presentations, I would recommend only having a voice presenting the topic at hand do as to avoid unwanted effects on learning. Thanks a million times for the video. Clear and to the point.
How do we choose which one of the indicators will be the fixed variable? I.e. in the example the path between the construct and X2 could be fixed to 1 aswell. Thank you!
"...the diagonal, which is shown in bold, indicates the variance, so the covariance of a variable with itself... gives us the variance of that variable. So covariance of a variance with itself is its variance." WHAT.
Great video! Thank you for posting the series! I have a couple of quick questions- at 34:55, why is there no variance associated with each of the square boxes (i.e., observed variables)? Also, are the parameter estimates for the errors always fixed to 1? Thank you very much!
Dear sir, your videos are simply awesome. You are explaining simply and understandably. Expecting more videos. Sir, I have one doubt too. When performing model analysis, I got an error like the model is unidentified and adds 9 constraints. How to resolve this issue sir? Could you please help me?
Hi at 14:47 point 3 I'm not sure what you mean by 'the sem' as S is the observed covariance matrix which you then compare against the model derived one (called capital sigma) by an minimalisation process.
Hello, will it be okay if someone can explain to me "what really is parameters?" I have ideas on it but, I can't really provide a definite explanation on it. Thank you so much. This will really help me writing my methodlogy for my undergraduate thesis.
Informative and useful lecture. However, I would be careful about repeating that social scientists aren't comfortable with mathematical models. This is an unnecessary generalisation. Yes, some are not comfortable, and some are comfortable with them. That's why there are scholars that specialise in quantitative measures while others use a qualitative approach.
Good day Prof Patrick. i am new to SEM. i am really confused why are constructs drawn using ellipse while making the measurement model may become rectangles and squares while making the structural model. please help me.
Well, yes. Useful background to an intimidating topic. But what do you actually DO? OK, what you actually do, sadly, at least in my case, is press some buttons in a statistical package and get lots of numbers out. But what do the numbers mean? Its surprisingly hard to find out and I suspect there is quite a lot of ""Ëmperor's new clothes"" going on out there Possibly this is covered in 3 to 6. But there is apparently no 3-6
at 1.5 speed it's still too slow and 70% of the words are ehs + Y = bX + e is not "a relationship" in general, its a very specific relationship, a linear one, i.e. it's a very very strong restriction
If anyone is searching all over RUclips for a video explaining SEM in a simplified yet concised manner, then this one's it 👌🏼👌🏼👌🏼👍🏼..worth 40 mins of your life.
This lecture series is amazing. I had to recently take a module on my PhD on Quantitative Methods, and as a first time learner of SEM, I find this video really helpful. Thank you very much Professor Sturgis.
not me out here learning about it while in high school 😭😭
@@hhhhhh359 Jesus, you learn this while in High School?
@@narendeepan yeah but by choice not bc its in the syllabus lol
@@hhhhhh359 Very good. You are smart.
Thanks a lot Prof Sturgis for these clear and informative lectures
I used to assume that SEM was quite difficult but this lecture series is truly enlightening. Thank you Professor Sturgis 😍
You really made my life easy Prof Patric. Thank you very much for this series of educating vedios.
Really easy to follow and clear, thank you very much for this video.
Thank you so much for posting this series, very helpful!!
Thank you, Dr. Sturgis! It's such a lucid lecture.
great lecture, very clear and well structured!
Your effort is appreciated
Thank you for the clear and concise explanation
The only productive critic that I would make is that while having a real person presenting the topic is perhaps more entertaining for the viewer, it adds a bit of extraneous load on them. As per Mayer's recommandations on multimedia presentations, I would recommend only having a voice presenting the topic at hand do as to avoid unwanted effects on learning. Thanks a million times for the video. Clear and to the point.
Disagree
Thanks a ton Prof Patrick 🙏 wonderful explanation
Very much grateful
Excellent!
Thanks for your sharing. It is very useful for researcher
How do we choose which one of the indicators will be the fixed variable? I.e. in the example the path between the construct and X2 could be fixed to 1 aswell. Thank you!
So easy to understand. I learned so much! Thank yoU!
Thank You Sir -Well Explained.If Possible kindly have a separate video on Observed variable with Example.
"...the diagonal, which is shown in bold, indicates the variance, so the covariance of a variable with itself... gives us the variance of that variable. So covariance of a variance with itself is its variance." WHAT.
Great video! Thank you for posting the series! I have a couple of quick questions- at 34:55, why is there no variance associated with each of the square boxes (i.e., observed variables)? Also, are the parameter estimates for the errors always fixed to 1? Thank you very much!
Very good presentation
Dear sir, your videos are simply awesome. You are explaining simply and understandably. Expecting more videos. Sir, I have one doubt too. When performing model analysis, I got an error like the model is unidentified and adds 9 constraints. How to resolve this issue sir? Could you please help me?
Thank you Prof Sturgis, quick check about the factor loading, why is it 2? ( 35.30 in the video)
Hi at 14:47 point 3 I'm not sure what you mean by 'the sem' as S is the observed covariance matrix which you then compare against the model derived one (called capital sigma) by an minimalisation process.
S as in Structure as in the variance/covariance matrix is the data structure of the model
here's a very BASIC question. who decides the relationships between latent variables- is it the researchers discretion ?
good professor
thank you very much. great help for me. it is clearly explaind.
Very good! Thanks for sharing!
Thank you very much
very good even as i need multiole viewing before i can really properly understand.
Hello, will it be okay if someone can explain to me "what really is parameters?" I have ideas on it but, I can't really provide a definite explanation on it. Thank you so much. This will really help me writing my methodlogy for my undergraduate thesis.
parameters summarize data for the population level, statistics summarize data for the given analyzed sample
👏👏👏
do correlated factor models explain unidimentionality if the 2 latent variables have strong correlations
Very helpful... thank youuu!
Thanks Prof Patrick!
Thank you
Thanks for this
Thanks you . your presentations are very formative.
Informative and useful lecture. However, I would be careful about repeating that social scientists aren't comfortable with mathematical models. This is an unnecessary generalisation. Yes, some are not comfortable, and some are comfortable with them. That's why there are scholars that specialise in quantitative measures while others use a qualitative approach.
very useful thank you, watching it in x1.25 =)
Thanks for the suggestion mate. it really helped!
Good day Prof Patrick.
i am new to SEM. i am really confused why are constructs drawn using ellipse while making the measurement model may become rectangles and squares while making the structural model. please help me.
Watch video 1 for more info
Everything is good. But the sections pertaining to Maximum Likelihood, Parameter Constraints, Nested Models and Model Fit needs more clarity.
Thanks Prof!
Well, yes. Useful background to an intimidating topic.
But what do you actually DO?
OK, what you actually do, sadly, at least in my case, is press some buttons in a statistical package and get lots of numbers out.
But what do the numbers mean?
Its surprisingly hard to find out and I suspect there is quite a lot of ""Ëmperor's new clothes"" going on out there
Possibly this is covered in 3 to 6.
But there is apparently no 3-6
Thanks! Part 3 seems to be missing though?
Hello, you can find all three parts and some supporting materials here www.ncrm.ac.uk/resources/online/SEM2016/
Thanks!
at 1.5 speed it's still too slow and 70% of the words are ehs + Y = bX + e is not "a relationship" in general, its a very specific relationship, a linear one, i.e. it's a very very strong restriction
Agree with you