The sample size was of three hundred respondents and the questionnaire has 45 questions.
Hi Tarika, please see the page below that explains the reason why:
http://www-01.ibm.com/support/docview.wss?uid=swg21476768
Dear Tarika,
I can't think why this is the case. What software package are you using? If you are using SPSS the KMO statistic (and Bartlett's test for sphericity) is one of the options on the Descriptives sub-dialog of the Factor Analysis dialog.
Depending upon the method you have chosen the factor extraction or rotation may not converge within the number of steps you have selected as a maximum. I understand that using Principal Component Analysis overcomes this problem. Non-convergence may be due to high levels of multicollinearity which should be assessed first by looking at the bivariate correlations between your items then removing one of each pair with excessively high correlation (e.g. higher than 0.8).
May be you lost the ticking the KMO and bartlett test of sph. test.
It may be because you forgot to mark the KMO and Bartlett Test in the descriptive submenu of Factor analysis menu. According to this warning, please try again.
@Nuri Dogan and Abdullah Faruk...thanks for reply ..I have ensured that I have tick marked all the submenus rightly
@ Peter Samuels..I'll try and find out correlation and will come back to you with results.
Hi Tarika;
Ok. I think you can use R code for this calculation. I attach a R code file for this purpose. And i wrote explanation of codes. You can use these codes to understand where the problem is. If you type in the error code given by R, it may be easier to comment.
Hi Tarika, please see the page below that explains the reason why:
http://www-01.ibm.com/support/docview.wss?uid=swg21476768
The link supplied above is indeed helpful.
Briefly: if the correlation matrix is nonpositive definite (some of the eigenvalues of the correlation matrix are not positive numbers) there will be a footnote to the correlation matrix that states "This matrix is not positive definite." and the KMO will not be displayed. This may happen if variables depend on other variables, such as one variable being a product or sum of other variables.
Dear Tarika,
I have experienced a similar problem, and I solved it as Kristen said. We just need to remove those variables which are derived from each other and just keep that one which is more biologically important.
Thanks Kirsten. My matrix shows "not positive definite". I now understand
You can run bi-variate correlation and identify highly significant variables. Retaining maximum independent variables and removing redundant information (removing highly significant variables) can help you out.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Values