View Full Version : Factor analysis
02-08-2006, 12:58 PM
Is anyone an expert on factor analysis?
02-08-2006, 01:13 PM
I haven't used it since grad school (and only a little bit there), but I could provide some basic guidance....
02-08-2006, 01:50 PM
Are there rules and regulations regarding the sample size (min) and the kind of scale?
02-08-2006, 02:08 PM
No - the larger the better. No restrictions on type of scale as far as I know.
02-08-2006, 02:38 PM
Can we use ordinal scale as continuous and do the correlation or regression analysis???
02-08-2006, 02:53 PM
Good question - it depends on who you ask and what their opinion is regarding ordinal vs. continuous scales.
Often you can, without concerns over getting misleading results - but take a close look at it to make sure your results "make sense."
02-09-2006, 08:00 AM
In SPSS, is there any distinction between exploratory and confirmatory analysis? Or do we just say we knew the factors already,,, which means we performed confirmatory analysis??? Thanks,
02-09-2006, 08:09 AM
In another question on this site, there's a link to a pdf file that walks you through how to do factor analysis in SPSS - you may want to check it out.
If you search the forums for terms like "SPSS" and "factor analysis" it should come up.
02-09-2006, 08:24 AM
Thank you, John, back to the previous question: I have only 40. So, that means that I can only use 4 or 5 items?? even though I am trying to get 4 or 5 factors from 8 items?? to correlate the factors with a physiological measure?
02-09-2006, 08:34 AM
I haven't seen your data set, and I'm not going to look at it, but I definitely would not arbitrarily pull items out just because I have a small sample size - in other words, how do you decide which items to pull out?
You could just do an exploratory correlational study to see which items appear to be correlated to each other - but again, the sample is small, and I wouldn't bet money on the reliability of your results.
So, you may only be able to reach extremely tentative conclusions, and you'll need to do another, larger study to confirm it.
02-09-2006, 08:38 AM
I have to let you know that a psychiatrist and a neurologist categorized the items already, and we are checking them to see if they really match. This is not the first study using the scale. They already know, although they did not run the factor analysis that time, that which items might go together. There are 26 items in the questionnaire, but only 8 items really correlated with the physiological measure. And only 8 items have the relevant factors in them. Thanks,
02-09-2006, 08:56 AM
It would be very helpful when you post a question regarding a study, that you provide us with as much information as possible. Many of your questions were quick, one-liners, and it's nearly impossible for us to give guidance on a study for which we have no knowledge.
02-10-2006, 08:03 AM
Hi John: Do people often acknowledge you in their papers for helping them???
02-10-2006, 08:17 AM
No, not that I'm aware of, anyway - and that's fine with me.
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.