This is a summary of Chapter XIII of “The Sensitive Nervous System” by David Butler.
Research has demonstrated that often evidenced-based medicine is low on the list for why clinicians choose a particular treatment. From an ethical standpoint, it is important to consider evidence. This chapter is very short so I will just provide the highlights that I got from it.
Appraising a New Theory or Approach
There are six criteria that a new theory should be evaluated by:
1) Support from anatomical and physiological evidence.
2) Designed for a specific population.
3) Studies from peer-reviewed journals.
4) Include a well-designed randomized controlled trial or single experiment.
5) Present potential side effects.
6) Proponents discuss and are open to limitations.
Here are some definitions of different ways research measures agreement.
– Cohen’s Kappa: Measures nominal data reliability.
- >0.75 is excellent agreement.
- 0.40-0.75 is fair to good.
- <0.40 is poor.
– Pearson product movement correlation: Measures interval/ratio data.
– ICC: Measures continuous data.
- The closer to 1, the better.
There are also many different validity types defined throughout this chapter. The first two are proven through logic and have the least evidence support.
– Construct Validity: Valid relative to a theoretical foundation.
– Content Validity: Can I use this measure to make an inference?
The next two are higher up on the evidence support hierarchy.
– Convergent Validity: The test shows a correlation between two variables.
– Discriminant Validity: The test shows a low correlation between two variables.
Lastly, these are criterion-based tests that infer similar results compared to an established test.
– Concurrent Validity: the compared tests are performed at the same time.
– Predictive Validity: The tests are compared at different dates.