Altair slc jacknife correctly computes standard error for complex survey designs
Too long to post on a list, see github
https://github.com/rogerjdeangelis/utl-altair-slc-jacknife-correctly-computes-standard-error-for-complex-survey-designs
Altair post
https://community.altair.com/discussion/49468/jackknife-tests/p1?tab=all
PROBLEM
Provide a better estimate of the standard error(SE) then weighted standard error
The jackknife standard error (7.28) is 76% larger than the naive
weighted standard error (4.12) because:
Within-cluster correlation reduces effective sample size
Unequal weights increase variance
Stratification was accounted for correctly
Design effect = (Jackknife SE / WEIGHTED SE)² = (7.28/4.12)² = 3.12
The jackknife correctly accounts for the complex survey design,
providing valid inference for the population parameter.
Jacknnife recognizes the true reduced indepenent degrees of freedom,
because of clustering. This inflates the SE.
There are many other uses for the jacknife.
CONTENTS
1 simple weighted SE wrong
2 jacknife SE wider SEDescription: Kindly add description here
Product/Topic Name : Kindly include the product name, version if applicable here. Also, Don't forget to tag with the product/topic related to your question.
*Note from Community Team - Once you receive a satisfactory answer on your question, kindly mark your answers as "Accepted" so it can help others with similar problems. Thanks.
