6
$\begingroup$

I'm trying to analyse if the fatality rate from my country (A third world country) vary significantly from the world's fatality rate.

So I'd basically have two samples, labeled (Philippines) and (World excluding the Philippines) then i can compute the fatality rate for the 2 groups.

Does Mcnemar's test apply here for me to check if fatality rate in the Philippines is higher, or do you have any suggestions? Thanks

$\endgroup$
3
  • $\begingroup$ What is the definition of fatality rate here? $\endgroup$
    – Eugen
    Mar 14 '20 at 5:31
  • $\begingroup$ It's defined as Deaths / Cases $\endgroup$ Mar 14 '20 at 5:34
  • $\begingroup$ Wikipedia's definition says it is the the same formula no matter which country. I also wouldn't directly associate a 3rd world country being threatened more by this virus, than any other countries, as long as there is no cure, the 3rd world countries actually should be threatened even less than others maybe. People in 3rd world countries don't travel that lot and they usually life in much smaller communities providing much less opportunities for the virus to spread. $\endgroup$
    – Eugen
    Mar 14 '20 at 5:41
4
$\begingroup$

It is not a case of paired nominal data. Hence, Mc Nemar's test can not be applied to check whether there is a higher fatality rate in Philippines ?. THE fatality rate is given for Philippines and world ( excluding Philippines ). As defined, it is expressed as proportion. Therefore, t-test/z-test shall be appropriate given that you meet other conditions such as sample-size.

$\endgroup$
2
  • $\begingroup$ Thanks appreciate it. $\endgroup$ Mar 15 '20 at 6:59
  • $\begingroup$ I would recommend an f-test as well $\endgroup$ Mar 19 '20 at 22:17
0
$\begingroup$

Adding to a previous answer of applying an f-test (diff distributions) and/or z-test (diff means), I think it would help if you would group into age brackets and applied your analysis there too. Perhaps a certain population fares poorly in comparison while another fares well.

I'd also suggest you factor some element of "reliability of testing" into your analysis, being that many people around the world (perhaps the Philippines) don't have access to testing or are under-reported in some ways. Such a factor could be a confounding var.

This is really important work! Keep it up!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.