Well, this is becoming interesting.
bblank correctly argues that multiple bad results will set the hook deeper on each iteration. It appears that bb's premise is based on statistical grounds alone. Never having been strong at statistics or other forms of lying -(lame joke)- I am more inclined to apply logic:
In the hypothetical of a marginal field test and an equal or worse result at the station house, you have given away the possible argument that the field unit, being subject to handling, etc, was out of calibration and false. If, however, you insist at the station house on the TWO blood tests at different labs, then you may buy increased quality of information for either conviction or exoneration. If both blood tests turn out clearly on the wrong side, then your goose is cooked and you will just have to deal with it. However, if one or both blood tests contradict the breath test, or if they are dramatically inconsistent with each other, then there's some helpful room for arguing that the facts are not strong enough to send you to the pokey.
Of course, if you're really ploshed, you probably won't remember all this anyway. In that case, the best strategy is to be someone else.