Can anyone give me a topic in statistics that might involve using randomization? just like the one sample test, anova, regression.. But this particular topics were already absorb in my class. So i want a new topic that my teacher wants me to discuss.
To tell whether there is any serial dependence in a series of data values such as temperatures, you can randomise the series at least 19 times and see if the lag-one correlation you get with the actual data is contained in the list of lag-one correlations you obtain from the randomised samples. This method can be extended for more than one lag.
The design-based (ie, probability based) approach to survey sampling is based on randomization, but I'm not sure that is within the scope of what you mean here.
Let us start from the beginning - what is the role of randomization or why is randomization needed? The 'statistical' reason for using randomization is to ensure valid (unbiased) inference that comes through randomly selecting a sample of study units from the target population. This may give you some clues to identify (practical) scenarios for "demonstrating" the value of "using" randomization.
Remember that while randomization provides unbiased estimates of the population parameter(s) of interest, it may however may not necessarily, practically speaking, provide a truly 'representative' sample of the target population! If you are a student of statistics, you can appreciate this point by considering the difference between "simple random sampling" and "stratified random sampling".
There is also a school of thought that considers randomization unnecessary, and think that a non-random sample carefully selected from the target population serves the purpose well! So... one topic that you like to explore is to compare these two scenarios taking one practical case - and demonstrate whether randomization, or a lack of it, is better.
Jane. check the following links to get some ideas
http://scholar.harvard.edu/files/kasy/files/experimentaldesign.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics