Hello,
I want to use logistic regression to estimate relationship between a dichotmous dependent variable and independent variables. To verify validity and reliability I will use explatory factor analysis, but a scale in the study needs to adapt in Turkish. As far as i know for adapting a scale, confirmatory factor analysis is required. Is it OK to use both CFA and EFA and then use logistic regression,
Thanx on advance,
best regards
Ceren Börüban, it seems like you already have an idea about the factor structure of your scale. You can thus subject your factor structure to a CFA right away without going via EFA. Moreover, you should be able to estimate both the CFA and logistic regression model in one omnibus model using a SEM software (e.g., lavaan in R).
Ceren Börüban, when you adapt scales from past studies, you may straight do CFA assuming that those scales are already validated. However, when you use the scales in a different context (e.g. in Turkish language in your context), IDEALLY you should start with EFA. To be more ideal, EFA and CFA should be undertaken on different samples, though unfortunately many studies do not follow this ideal practice.
All the best,
Khan
If you are using SPSS, the add-on module AMOS will handle both CFA and SEM.
You might have a look at the insights in Barbara's (2016) third-edition book, Structural equation modeling with AMOS: Basic concepts, applications, and programming, fully cited below.
Byrne, B. M. (2016). Structural equation modeling with AMOS: Basic concepts, applications, and programming (3rd ed.). Routledge. https://www.routledge.com/Structural-Equation-Modeling-With-AMOS-Basic-Concepts-Applications-and/Byrne/p/book/9781138797031
Good luck,
It's much better to use EFA and CFA in combination for the adapted scales that are not yet univesely validated.
Only CFA can be applied But using more techniques to validate the data is always prefferable. Therefore you can use EFA and CFA.
It can be. Usually, EFA is conducted before CFA. However, it depends. if you search for a new factor structure, EFA should be done and then validate the explored factor structure identified. However, you must ensure that the sample size is large enough to allow for split sampling (half for EFA and half for CFA).
Hello there,
CFA is the improved version of EFA with the introduction of restrictions in determining the goodness of fit.
EFA can be done as a supplementary for CFA in ensuring validity and reliability of the instrument. But in the end, CFA will measure the items in the instrument and will eliminate the items that did not fit in the model.
Thus, it goes back to your objective and what actually that you would like to get from your instrument. (In some cases it is crucial to run both analyses).
[[could you please help me]]
I have 8 independent variables and one dependent variable for study predicting accepetance of the robot in delivery . if i made the independent variables by 5 likert scales and each variable has 4 or 5 questions. is it possible to use logestic regression and how ?
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publications