Hello Eve,Is highly likely that you might have an eigenvalue which is negative, causing this error. Generally, this will make the covariance matrix of a multivariate normal distribution not to be a positive semi-definite matrix.
Depending on the software you are using, the negative eigenvalue(s) can be set (forced) to zero(0) if is not far from zero and then a new covariance matrix generated from that for the computation process to continue. Hope this helps!
Kind regards
**********************************************************************
Dr. Justice Moses K. AhetoPhD Medicine (Statistics & Epidemiology).
MSc (Statistics with Applications in Medicine).
BSc (Statistics).
HND (Statistics).
International Research Award: http://howafrica.com/ghanaian-phd-student-wins-top-uk-prize/ http://www.graphic.com.gh/news/general-news/ghanaian-phd-candidate-wins-top-uk-prize.html
Professional Interview with BBC World Service: http://www.bbc.co.uk/programmes/p04knlvj
International Recognition: https://www.theguardian.com/science/blog/2017/oct/02/why-were-adding-black-mathematician-month-to-our-calendars or https://www.lms.ac.uk/news-entry/03102017-1422/october-black-mathematician-month. Professional websites: https://www.linkedin.com/in/justice-moses-kwaku-aheto-a7338647
**********************************************************************************Lecturer and ResearcherDepartment of BiostatisticsSchool of Public HealthCollege of Health SciencesUniversity of GhanaP. O. Box LG13
Accra, Ghana
E-mail: [log in to unmask] [log in to unmask]: justice.aheto
Mobile: +233(0)242527292
***************************************************************************************
On Friday, October 5, 2018, 9:46:54 PM GMT+1, Eve Namisango <[log in to unmask]> wrote:
Hi allstat, I am running factor analysis for categorical variables and getting this error message [ mysigma not positive (semi)definite]
Does anyone have thoughts on what I have to do to move forward?
Eve
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|