Dear Colleagues:
Here is a question on Bootstrapping. You need not be a bootstrapping expert to give an opinion. Does bootstrapping require that you use the same statistic as the original point estimate? Silly Looking Example: The variance of the exponential distribution is the same as the mean. You quote your point estimate as the mean but bootstrap on the variance. You get a correct confidence interval, but does this violate 2 and/or 3 below? I would appreciate your answer and will share responses in mid March.
How Bootstrapping Works
- Original Sample: Start with a representative sample of size N
from the population.
- Resampling: Draw a new sample ("bootstrap sample") of size N
from the original data with replacement.
This means some data points may appear multiple times, while others not at all.
- Statistic Calculation: Calculate the statistic of interest (e.g., median) for this resample.
- Repetition: Repeat steps 2 and 3 a large number of times
- Distribution: Analyze the distribution of these bootstrap statistics to estimate
standard errors and construct confidence intervals.
Jon
------------------------------
Jonathan Shuster
------------------------------