Jessica, my memory may have modified a little what I actually did read but not by much I hope. The article is by Steven Foreman (The significance of turing passive into active in control mastery theory) in The Journal of Psychotherapy Practice and Researach, 1996, 5, 106-121.
Before citing the part where I inferred it is better to err by assuming transference testing if the therapist is baffled, I want to first say that what you explained, Jessica, is also referred to in Foreman's article and with much emphasis. Thus, Foreman wrote (p. 113) that: "Control mastery theory says that with few execptions, passive-into-active testing in the therapy should not be stopped or even discouraged." (Two important exceptions mentioned were when there is danger of physical harm and when the patient feels guilty about such testing. With the latter exception it was not clear that the therapist would be encouraging testing to stop necessarily. The point was that the therapist may help the patient feel less guilty or the patient may want to stop if the patient is readily informed of the passive-to-active interpreted meaning of his/her behavior.)
As I noted in my posting above, Foreman also advised (p. 113) that passive-to-active interpretations can be experienced by the patient as criticisms or rebuke. Foreman: "Often when therapists interpret patients' turning of passive into active, their interpretations truly are criticisms or rebukes--and no wonder, since therapists expereicne this behavior as quite toxic and provocative. Often the interpretation is an attempt to stop the patient from being so unreasonable or obnoxious by pleading with, instruction, or scolding the patient into more pleasant behaviors." (p. 113)
Then (p.120) Foreman writes just before his conclusion: "In general,it is recommended that in mixed or confusing tests, the therpist should err on the side of failing a passive-to-active test rather than a transference test. Generally the patient makes himself more vulnerable in transference testing than when turning passive into active. If the therapist fails a passive-to-active test by appearing weak, for example, the patient will not be too damaged, but will become anxious and test again. In transference testing, if the therapist repeats the transference trauma often enough the patient will be damaged again as she was in the original family trauma." It looks as if I interpreted this passage to mean that more interference with a patient's progress may be done to a patient by a therapist's failing a transference test than by failing a passive-to-active test. Thus I concluded that when a therapist is unable to distinguish which type of testing is taking place it is better to risk err by relying on an assumption of transference testing "until further notice."
Combined with the above caution that therapists may be unwittingly (or wittingly) tempted to be unhelpfully critical in interpreting passive-to-active testing, there is all the more reason to err on the side of assuming a transference test when is doubt about which type of test one is facing.
However, what I have tried to emphasize applies only to situations where the therapist is in great doubt about what kind of test s/he is presented with. My idea about assuming transference testing insn't relevant to other situations, and such an assumption would be misapplied if a therapist were to try to make his or her work less distressing by too easily overlooking the presence and value of active passive-to-active testing in the course of a patient's progress. Similarly, Foreman is careful to end his very useful article with a reminder that the therapeutic alliance, for patients who must rely on pasive-to-active testing, will consist of what is essentially a stormy time for therapists. With some patients there IS a therapeutic alliance but it doen't look like it if one thinks that an alliance exists only if the consulting-room atmosphere is frequently free of affective pyrotechnics in action. Thus the concept of a therapeutic alliance must be extended to include patient-therapist relationships characterized by stormy passive-to-active testing for those patients who "do therapy by action," that is, whose best available method of communication is enactment (adolescents, children, borderlines, and---possibly?? what do you all think?--MPD/DID patients or those with amy other condition related to severe trauma).