20th June 2013 at 4:47 pm #1519Hermes MercuryMember
I am trying to get my head around a problem but I am finding it difficult.
As an example, let’s say that the literature shows that a treatment (e.g., cognitive behavioural therapy; CBT) is shown to have a large effect size compared to a control for a disorder (e.g., depression).
If I hypothetically want to argue that although the effect size for CBT, compared to a wait list control, is large, the patients are still more depressed than the general population meaning that although CBT improve symptoms, it does not do this to an adequate extent so that symptoms of depression no longer get in the way of daily functioning, and is thus a weak treatment.
My questions are (1) whether or not it is breaking any “research methods logic” to compare depressed patients that completed CBT to the general population, and (2) is would it be a sound criticism of a treatment to say “yes it improves symptoms better than other treatments BUT the symptoms of depression still (a) get in the way of daily functioning and (b) they are more present and to a higher degree than in the general population”.
The reason I am asking this is I want to criticise a current treatment and suggest an alternative (not CBT for depression, that was just an example) but this current treatment is shown to be effective by the literature. However, I would hypothesise that the alternative treatment would have a larger effect size when compared to other treatments, would reduce symptoms more than current treatments and would result in patients being more similar to the general population in terms of specific outcome measures.
Many thanks for the help21st June 2013 at 3:22 pm #1522Dave CollingridgeParticipant
Interesting study objectives. I see nothing wrong with comparing CBT patients with the general population. In fact such comparisons are ideal for determining whether the therapy restores people to ‘normal’ mood states. With the average outcomes values for norms on hand (from literature), you could compare CBT and normals with single sample t-tests. You just need to decide which outcomes you will compare.
It would be interesting to compare outcomes for both treatments. Collect data on outcomes from both groups and use appropriate statistics to compare. If you could include normals, it would make for a nice 3 factor (group) comparison and you could skip single sample t-tests above.23rd June 2013 at 8:03 pm #1521Hermes MercuryMember
Thanks very much for taking the time to reply.
I’m glad that you answered this question because as in research papers I usually come across only clinical controls and clinical treatment groups I felt that including healthy controls might be a problem but thanks very much for confirming that it’s a feasible idea. I suppose that as long as an idea, theory, argument and so forth is logical, there is little reason why it being unconventional would make it wrong or any less logical.
Nice idea also with the 3 factor (group) comparison, that would indeed be an interesting analysis to conduct.
P.S. I’m a new member of this forum but I am definitely going to be checking it out more. A forum where it’s possible to discuss research with others is superb.24th June 2013 at 4:07 pm #1520BIRHANU W/YOHANNES SEYOUMParticipant
you can use t -test.
- You must be logged in to reply to this topic.