Recently, I came across the issue of stochastic variation, in which the percentage of response in every items in a questionnaire administered was between certain range. Could these be due to the fact that the questionnaire used is too common, in which the respondents had prior knowledge to the scale? The issue of stochastic variation remains unclear. How to reduce or limit such phenomenon during analyses if possible?
Dear Kurubaran, I have been working for a while on the subject you are bringing in. Personally I have a rather radical point of view. I apologize in advance if my answer may not help you so much.
The satndpoint from which I tackle the question are the sciences of complexity (in general). One of the most salient features of what a complex system is is characterized by randomness, indeed; its stochastic character, if you wish.
Well, the standard way of coping with randomnness is via analytical tools. However, properly understood, a complex system can not be parametrize, for parametrization kills complexity. Analytical tools are highly succesusful (for practical reasons), but they just linearize non-linear behaviors.
Statistical anaylisis, Lyapunov constant, parametrization, and many other analytical tools are broadly taught and worked out. By them however, we just displace complexity - we do not solve it in its radical challenge.
New tools and rods are to be developped that do not kill complexity, but rather help us to harness from complexity.
Thank you Professor for the explanation. Despite analytical tools to linearize non-linear behaviors, can the stochastic phenomenon be controlled by it's demographic effects in a population based study? As such would it be able to control such variability if sample size is small or vice versa?
From the technical point of via the answer is: yes. You can control the stochastic phenomenon in a population based study. The question is rather, why should you control it? AsI said, you can control, hence linearize stochasticity. But then one is tricking oneself.
Simulation (not modeling) helps understand stochastic, i.e. random processes or structures. It all depends on the researchers purpose.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis