Also the patient wanting to please you. Patients can feel coerced into participating. It is good to have someone other than the therapist obtain consent,

Dear Jan, 

here are some examples: 

Article Family therapists as clients in therapy

Article The Nature of Ongoing Dance Engagement and Identity as a Dan...

Article Researching Ones Own Clinical Practice: Managing Multiple Ro...

I used data obtained by others care professionals at the same time, as a mesure of other related variables of the evolution; for example changes in medical prescriptions. These care professional was not involved in the research.

You could consult my paper "Análisis de la eficacia de un tratamiento grupal cognitivo-conductual en sujetos con somatizaciones" (in spanish) or "Effectiveness of cognitive-behavioral group therapy in patients with anxiety disorders" (in english)

As practitioners we are both researchers as well as therapists , it is always  preferreable to use the consent of other coworkers to understand the degree to which the others agree with what your findings are especially when it comes to qualitative studies it is advisable to have other researchers as a part of it to remove biases . As in thematic analysis or even phenomonological enquiry , it is best to allow others to agree upon the themes arrived at to rule out any bias on the part of the therapist cum researcher.I do take the help of a psychiatrist if it requires so. 

An analyst analyses her own work in analysing. It is psychoanalytical research ...is there anything which is not subjective..

The therapist's aim is to improve the lot of the patient using an intervention  which he considers worthwhile. He has spent considerable time training for it and considers himself skilled in its application.

The researcher's aim is to discover whether that intervention is better or worse than an equally plausible one.

The two aims are at odds.

If therapist and investigator are the same person, the patient/research participant has not been randomly selected for the study; neither patient nor investigator/therapist  is blind to the purpose of the study; both participant and researcher are heavily invested  in showing the superiority of the therapist's intervention and in minimizing  any side effects or downsides.

How can one doubt the result of such a study?

Not being blind to the status of the patient and demand characteristics from the patient. There are also a host of other biases that can come up - confirmation bias, illusory

correlations, neglect of base rates, and availability heuristics, to

name a few (although these can happen irrespective of the dual role of therapist and researcher). 

Piggy-backing on David's input, if we think of a therapy session in terms of an experiment and individual therapists' model of approach as levels of the independent variable (which is actually the reality of psychotherapy unless the client does not wish his/her case information released for studies), then there is an astronomical number of confounding variables to control for: Rather than the patient's behavior changing in accordance to the model of approach--which we'd hope for, it can change in accordance to unconscious non-verbal and para-verbal signals from the therapist, such as movement of a pen towards a piece of paper in waiting for an answer (Rosenthal Effect) or in accordance to simply being observed (Hawthorne Effect). What's worse is that these variable's cannot be teased out of the therapist-client relationship. However, there is ongoing development in subtle ways to control for the therapist-client relationship.

The 6-Item Working Alliance Inventory (Short Form Revised) that is administered after therapy sessions has the potential to assign quantitative values to components of the therapeutic relationship (therapist-client agreement on goals, tasks, and emotional bond). With psychometrics that control for these kinds of variables in measurement of client outcomes, psychologists may be able to determine the effects of variables in the treatment as less confounded by relationship dynamics.

Paula Schnurr wrote a great review and summarized the most important biases influencing psychotherapy outcome research.

Schnurr, P. P. (2007). The rocks and hard places in psychotherapy outcome research. Journal of traumatic stress, 20(5), 779-792.

Respected sir,

I agree with you but it can not become a problem.Biases happens in single-case experimental designs, where the researcher and therapist are same person. It introduce the biases such as unnecessary emotional thinking, tendency to good result and conclusions,sometimes over activity or practices,uncontrolled experimental situation,believe system, safety feelings,lack of transparency in experiment,contextual effect may be becomes a biases in this method.

Another good paper demonstrating the impact of researcher allegiance on psychotherapy outcome.

Article Researcher allegiance in psychotherapy outcome research: An ...