factor analysis

[lynnar]

HI,

I conducted interviews with 30 people on their perceptions towards Device X. Based on previous literature, the perceptions woud be categorised under 3 factors: Factor A, factor B and Factor C.

I've transcribed all the interviews data and roughly I found some additional factors, namely Factor D and Factor E.

My question is, how do I analysed these and confirm on these factors? I know I could use confirmatory Factor analysis (CFA), but I don;t know how I could do it. The materials I found over the internet mostly explain on the statistical formula. I wanted to know how CFA works (not in statistical explanation).

I would really appreciate any reply..

Thanks a lot!!

 

[lynnar]

I would really appreciate if soemone could help me..most of the books that I've read talked about the results of scale (e.g: from scale 1 to5, how do you rate your satisfaction using this device?). My data are all qualitative data where the users provide their feedback in open ended form. Can someone please let me know how am i supposed to analyse this?

Thanks a lot!!

 

[noetsi]

Well the first thing is that thirty cases will give you really weak power. So your hypothesis tests if you do any will be very weak.

I might be able to answer questions about CFA but I need to know what you want to know specifically. You don't confirm the existence of a factor with CFA you test theory about how factors work through observed data. The existence of factors is suggested by EFA and theory and proven though tests such as discriminant validity. Supported by theory developement.

CFA (which is a form of structural equation modeling) is very complex and requires very high sample sizes. You should not use it (at all) with just 30 cases. It is not a small sample size method - ever.

 

[lynnar]

HI..i think i might have got it wrong...what I am suggetsing in this post id to use CFA for qualitative data...i just relaised that CFA could only be used for quantitative data (usually likert scale). I might have understood this wrongly.

Thanks fr your reply anywya.