Here is some language on this topic from an ASCE/SCEC document which was created to be a guideline for landslide analysis in Southern California:

"The methods of Morgenstern and Price, Spencer, Sarma, Taylor, and Janbu's generalized
procedure of slices satisfy all conditions of equilibrium and involve reasonable assumptions.
Bishop's modified method does not satisfy all conditions of equilibrium, but is as accurate as
methods that do, provided it is used only for circular surfaces. Duncan (1996) has found all of
these methods to provide answers within 5% of each other.

Considering the foregoing statements regarding accuracy, the methods of Morgenstern and Price,
Spencer, Sarma, and Janbu's generalized procedure of slices probably will yield reasonable
estimates of the factor of safety for failure surfaces of any shape. However, because of the
difficulty associated with selecting an appropriate force function for use with the Morgenstern
and Price, and Sarma methods, and the frequent numerical instability problems associated with
Janbu's generalized procedure, those methods may not be suitable for general engineering
practice. As a result, the Committee recommends that Spencer's method be used for analyses of
failure surfaces of any shape. In addition, we also recommend that the Taylor and Bishop
modified methods be allowed for the analysis of circular failure surfaces. If a stability analysis
has been performed using a method other than the Spencer, Taylor, or Bishop methods, it is
recommended that the factors of safety for critical surfaces be checked using one of those three
methods."

You can download the doc from here:
http://www.scec.org/resources/catalog/LandslideProceduresJune02.pdf