Unfortunately, this isn't something you can do in SPSS. Latent Class Analysis is definitely where I would start, which would allow you to identify classes of "opinion profiles" for respondents, based on patterns of responses, and then determine if owners of each type of phone are most likely to be in particular profiles.etc. PS: Sarah, might the same be happening to you? In your case, I tried running your analysis without that line of code (without selecting that subset of cases) and your sample size is 213, which is enough to allow the model to run correctly (the output then produces KMO and Bartlett's). There is no clear-cut guide for how many people are required per variable (although i've heard anywhere from 5 to 20 used as a rule of thumb), but you should definitely never have more variables than people in your analysis. The problem is: doing that shrinks your sample size down to only 30 people, which is far too few people when you have approximately 40 variables to analyze. This command selects your analysis to include only individuals that respond with "1" for question "Q2". The tenth line of is where problems arise: " /SELECT=Q2(1)" Below is your syntax:ĭATASET ACTIVATE DataSet1. I think your problems stem from your sample size, specifically when you select your sample by people that answer Q2 with "1" (which is what was done in your analysis). Since you sent me your output and data, I have a bit more insight into what is likely going on with your situation. If anyone can help I would be eternally grateful!! I can send you the results in SPSS if this helps to demonstrate what I have tried to explain above. I have used PCA to analysis the results (I know that this isnt the ideal FA test to use but this is the only one I can get any results out of even if they are wrong!). 9 and no variables that correlate higher than. I have also checked the correlation and there are none that correlate at higher than. I have gone through my results and none of the participants seem to have answered in a weird way (e.g. 00000 and that the matrix is not positive definite. I don't know why this is happening and think it may be something to do with multicollinarity as at the end of the correlation matrix it says the derminant is. significant section of the correlation matrix and the KMO, barlets). However there is something wrong with my data so that when I try to analysis it a lot of the tables that are meant to be given in the output don't appear (e.g. EEK! I think I have vaguely got my head round factor analysis (I am not mathsy at all!) and am attempting to use it in order to validate a questionnaire I have designed. My supervisor has left the country and only gets back 10 days before my final hand in and I have no results. I am struggling with my dissertation project for my masters.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |