Understanding Society User Support

Dear members of the support team,

I am using the UKHLS to analyse the well-being of self-employed individuals.
For my dataset, I need to filter for self-employed people, and I am uncertain whether to use jbsemp or jbstat.
I have noticed that filtering on jbsemp (self-employed) results in some cases where jbstat is recorded as “retired”, “paid employment”, “family care”, etc. In my current sample, this mismatch occurs in approximately 10% of observations.

Could you kindly clarify how I can interpret cases where jbsemp = self-employed but jbstat is not “self-employed”?
For exampple, if jbstat is "retired" but jbsemp is "self-employed", does this indicate that the person is doing some form of self-employed work while in retirement, or does it reflect their previous employment status before retiring?
For research on currently active self-employed individuals, would it be more suitable to filter for both variables or only one of them?

I would appreciate your support.
Thank you & best regards!