Dear Ms. Beare,
The reason for the discrepancy in these admixture models is that the
categories are too broad. Each of the so-called ethnicities is a mix of
many other ethnicities.
To help you understand your results better may I suggest that you upload
your raw files to gedmatch.com (free) and then run them on the
'Admixture' analysis using the Eurogenes K36 model. This will show you
more nuanced results according to 36 ethnic categories.
Have fun and don't expect black or white results.