study so that you can be SURE that what you intended caused any change. It avoids someone asking "Might X have caused the change ?", and you having to reply "Maybe" or "Possibly". If so, you don't know what caused the change and the study hasn't explained anything. It has been "confounded" by X. Case Studies
A very common way of reporting the effect
of treatments on patients.
But why might case study reports NOT be
acceptable evidence that a treatment works? 2 Problems With Case Studies
1 The patient might be getting better already
2 Illness might be in remission 3 Another factor might cause improvement 4 Patient might feel obliged to report improvement 5 Therapist expectation might bias measurement
Summary: Improvements cannot be attributed to
treatments with certainty. Clinical evaluations are designed to avoid these problems. 3 Types of Confounds
1 FACTORS OUTSIDE THE STUDY
1.1 Were they getting better anyway?
1.2 Something else could have caused the change
2 FACTORS INSIDE THE STUDY
2.1 Changing several things at once
2.2 Patients' expectations
2.3 Clinicians' expectations and measurement bias
1 FACTORS OUTSIDE THE STUDY
1.1 Were they getting better anyway?
Solution: Have a baseline period
Improvement shown, but they might have being getting better already. There’s a baseline here, but something else could have caused the change 1 FACTORS OUTSIDE THE STUDY
1.2 Something else could have caused the
change
Solution 1: Introduce intervention at different
times 1 FACTORS OUTSIDE THE STUDY
1.2 Something else could have caused the
change
Solution 2: (many patients) Have a
comparison control group. measure 2 FACTORS INSIDE THE STUDY
2.1 Changing several things at the same
time
Solution: Avoid or divide patients into
different groups 2 FACTORS INSIDE THE STUDY
2.2 Patients' expectations
Solution: Have comparison groups with a
"sham" treatment; have a cover story or "blind”. 2 FACTORS INSIDE THE STUDY
2.3 Clinicians' expectations
Solution: Have a similar "blind" to stop
unconscious influences. 2 FACTORS INSIDE THE STUDY
2.4 Clinicians' measurement bias
Solution: Ensure that the therapist does not
take the measures by using an objective measure or an independent “blinded” observer 5 Examples Of Research Designs CONDITIONED PHARMACOTHERAPY Averting Epileptic Aura
Example Design RULE NUMBER ONE (Yet Again)
If you fail to plan,
you are planning to fail GROUP TASK
1. Select a condition and treatment
you’d like to evaluate 2. Sketch out an Introduction (with mock references) and Aims 3. Select a design 4. Prepare the Procedure (what each patient will go through)