I had prepared a 155 item questionnaire and had subjected the data collected with it to factor analysis (principle component analysis with orthogonal rotation because I wanted independent factors). The result reflected 75 items and 7 factors (initially they were 18 factors).
Thank you for the response. Yes, it is helpful. I am doing this model for the first time and no such work is done up till now. I was confused because few of my teachers were recommending Model Fit, even if this model is prepared for the first time. I
I have another doubt if it is ok for you to help. I want to propose a model for personality. So preparing or modifying questionnaire or scale after every factor analysis is mandatory?
I prepared a scale of 155 item and 19 factors, factor analysed and got 36 items and 7 factors. now i have collected data on 36 item scale and factor analysed. a different set of results are obtained. I collected data third time with modifying same scale with an input of 150 item, but yet to anayse.
hence my questions stays, Do I need to modify scale everytime I propose the model?
Also you need to take care of latent variable (construct) while doing EFA coz sometimes its absurdity of observed data for a variable which leads to a wrong association with a Factor.
E.g.- You have an output of 2 factors F1 and F2, formed of (v1,v2,v3,v4) and (V5,v6,v7) respectively but in reality the v2 should reflected by F2, because of data points(sometimes not captured well) this things may appear, so you have to take a rivew of all observed variables while finally analysing your path model with Factors.
& yes for a detailed Exploratory model analysis go for SEM.
You can use SmartPLS for your Phychometrics Analysis, Its very easy & user friendly to use, AMOS might be a bit complicated in start for a non-statistician. But there are certain restrictions too.
hope it helps.
Rgds,
Abhay
Thank you, Abhay. Your suggestions are very helpful. I would like to ask another question. If my item loading in one factor shows a magnitude of 0.9 then it is considered high loading, what if the loading goes to 1.08?
how shall I interpret such magnitude?
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS