JASP 0.13.1 Tutorial: Exploratory Factor Analysis (EFA) (Episode 20)
HTML-код
- Опубликовано: 23 сен 2020
- In this JASP tutorial, I go through an Exploratory Factor Analysis (EFA). I use early preliminary data to explore features including Rotation, Factor loadings, Eigenvalues, and how to read the tables in the results.
NOTE: This tutorial uses the new preview/beta build of 0.13.1. This build contains slightly more functions/features than the previous builds used for tutorials on this channel, but it is functionally the same for the purposes of this tutorial.
Find me on Twitter: / profaswan
Go to my website: swanpsych.com
Thank you for all your JASP tutorials! I'm using JASP for my bachelor thesis in legal psychology as an affordable alternative to SPSS and wouldn't have been able to learn it that quickly without your videos (and time is of the essence). So your efforts are very much appreciated.
Great sir 👍 am waiting for this
Thanks
absolute legend
Helpful. Thank you.
Good, it works for me. If the original dataset is available to reproduce outputs, please let us know.
tq sir.. very usefull
Thank you for your videos Alexander! Could you please make a tutorial about how to perform Confirmatory Factor Analysis in JASP using a grouping variable, for example Gender as a grouping variable. Thank you!
Sure. I'm not sure how well that will work, but I can give it a shot. I am not super well-versed in all the measurements/fit indices for CFA, but I'll see what JASP has to offer on this front.
Hey there! Thank you so much - this was very helpful!
Could you explain to me how to calculate a item-total-correlation in JASP?
I'm not sure where that function is in JASP. I'd have to look.
Thank you for the tutorial. Is there a tutorial that explains Regularized Exploratory Factor Analysis (REFA)? Thank You.
I’m not familiar with that technique, so I am unsure if it is possible in JASP
Hi, and thank you for the informative video. I was wondering what the scores under "cumulative" under "factor characteristics" tells us? If I understood it correctly the proportion var. tells us how much of the modell is explained by the factors. But I am not sure about the cumulative. Anyway, thank you for the video
Cumulative usually sums to 100%
good day, alexander, and thank you for the tutorial! if i'm allowed to ask a question, i have one question. if i test my data from multidimensional inventory with EFA, should i test the data's reliability per its dimension or with the factor from the EFA result? thank you in advances!
Always test reliability within dimensions before and after running your factor analysis.
@@AlexanderSwan thank you for your answer, it helps me a lot!
hi, i need major help with how to examine the dimensionality of the data (bear in mind that although the intention was to develop two scales, it is possible that the 30 items measure more than two major traits or dimensions) how do i do this
This isn’t something I can help with without looking at your data and output, sorry
Great Teaching,!.. ❤️May I ask...Does Uniqueness in Factors matter???.. I've seen your uniqueness values abOve 0.4... Do I have to remove all the questions that Have below 0.4 value in the Uniqueness as I do The Exploratory Factor Analysis???.. Thank you💙
This is a convention among statisticians -- .40 is seen as a good level for an item to be a good contribution to the factor. Items below that value don't contribute much to the factor, so they can likely be removed. If you don't remove them, nothing much changes, but you might get questions as to why they're still there
@@AlexanderSwan Thanks❤️
what to do if the variables are originally nominal?? and it cant be changed into scale :(
While you *can* perform an FA with nominal data, it won't look good. This is because an FA is a matrix of Pearson correlations, which you cannot use nominal data with, anyway.
Thus, I don't recommend. If you are looking for nominal correlations, you want polychoric, but JASP can't do that.
@@AlexanderSwan oh gosh I thought I wouldnt get a reply, thank you so much for this and you are an actual blessing
Sir I want to ask. If my thesis proposal is ready, is the first step in this efa distributing the questionnaire to test its validity? Or how?
Pak saya mau bertanya. Kalau proposal skripsi aaya sudah jadi, langkah awalnya dalam efa ini menyebarkan kuesioner dulu kah untuk uji validitasnya? Atau bagaimana?
I'd say the only way to explore your questionnaire is to collect some data! Collect as much as you can for a stronger factor analysis
@@AlexanderSwan
sir want to ask again
1). If in efa it means that the indicator is obtained after we distribute the questionnaire, right? For example, there is a brand image variable = bi1, bi2, b3, b4 product quality=pq1,pq2,pq3,pq4 and the indicators are obtained from the results of distributing questionnaires. So basically, the indicators are not obtained from the opinion of experts?
2). Is there a minimum limit for the number of variables and indicators used in the EFA?
@@sembilanduabelas680 apologies, but this is not something I can really delve into here on YT. I don't fully understand your question without knowing background. There is a theoretical limit on variables in an EFA, but I'm sure what you have is fine.
@@AlexanderSwan ok sir, im sorry if my question is a little bit OOT.
Okey i'll try to learn more about EFA, coz actually im not english native (i'm from Indonesia which is the source or book about efa/cfa is not many in Bahasa) so i'm little bit have a problem to understand with english language.
The lastly, thx u so much again Sir
How do we interpret results tho
That’d be in another video made by someone else. I do minimal interpretation, rather here I just show users how to use the module.
thanks! where is the "explained variance"?
"Explained variance" is not an output but rather how you interpret correlations and factor loadings. So it's just something I say when talking about the results!
@@AlexanderSwan só, is not possible to get this information?
@@icarocosta4302 I'm not following your question.
@@AlexanderSwan that % of explained varience. Is not possible to get it from jasp?
@@icarocosta4302 oh I see. It is in the output as the load coefficients. Then you just use that to discuss variance