suspected of being
problematic, one is a
common
but
questionable
model, and an
adjustment
may
be
required
to address all
the nonsignificant
differences—
how
they
nonetheless constitute
important arbitrary
criterions
for
equivalence

the significance test
based on
observational
data
is
susceptible to (errors
of) interpretation
over the
question
at issue
namely, do
case differences
arise
because of
exposure
to a comparatively
small sample
or
because
of
another variable?

Exposure can be
only mediated
by
crude
estimates
and so may be
misleading
during
the
forming
of the hypothesized
model of one
that describes
the
association
between exposure,
bias, and
the variables,
and
reconciles
difference
with equivalence
significantly.

The model provides
little information
that is
incontrovertible
but
the results suggest if
adjustment for the variable
makes no
substantive
difference
ignore it

but if your knowledge
indicates the
adjusted
variable to
be preferable

then prefer it