I read that I have to use software such as AMOS or LISREL but being a mac user, I don't have access to any of those. I need to run CFAs on previously validated questionnaires.
I read a recommendation on another similar question that using EFA on each dimension of a scale (e.g. parenting scale) is acceptable. It doesn't seem to be widely practiced though. Is simply running an EFA as a way of "confirming" factor structure acceptable?
I am already running Cronbach's for reliability; are there alternative psychometric tests (other than CFA) that can be conducted if i'm using those questionnaires for a new study?
Thank you.
No, you can not run CFA in SPSS. You have to use LISREL, AMOS, EQS, or MPLUS to perform such analysis. The AMOS is the most user-friendly software of all; EQS is the second one. I'm not sure which softwares you can use on Mac, but if you run SPSS on your OS probably you can also run AMOS. You can use EFA to analyse construct validity, but the CFA is more robust. EFA is not "confirmatory". And using EFA on each dimension is not widely practice at all.
And R software, but AMOS still is the most user-friendly software of all and EQS the second one.
Dear Claudia
Do you mean that I have to refer to previous CFAs conducted by the scale authors (or other researchers who used the scale)? To my knowledge, they only reported EFA values with no cross loadings. Is it possible to conduct a CFA based on the values they reported in their EFA?
Thank you!
Hello Zw,
Unfortunately, the AMOS add-on for SPSS is only available for windows operating systems. Likewise, EQS and LISREL are also only available for windows systems. As Claudia Lopes indicated earlier, both MPLUS and R are available for Mac OS. Within the R package, the SEM or LAVAAN libraries are more than capable of handling CFA.
Alternatives to CFA (the other part of your question) for validity evidence could include: (a) multitrait, multimethod approach (by Campbell and Fiske, 1959)--here's a link that will give a brief introduction: http://www.learningpsychology.net/Multitrait-Multimethod-Analysis-(construct-validity).html; or (b) Procrustes rotation of one factor solution to compare with another factor solution (and subsequent metrics for similarity/dissimilarity of solutions)--here's a link for an R library function to do just that: http://cc.oulu.fi/~jarioksa/softhelp/vegan/html/procrustes.html
Neither approach is as straightforward as CFA, but can be helpful for understanding the nature of a measure (or set of measures).
Of course, you can purchase a windows license and install it on a mac, here's a link that shows a couple of ways to do so: http://www.macworld.co.uk/how-to/mac/how-run-windows-on-mac-vmware-boot-camp-virtualbox-3497251/
Good luck with your work!
No you can't run it in SPSS. Softwares like R which is command base AND AMOS can be used for CFA