When N is less than 5......?

[Raouf Alami]

Hello everybody,
  I would like to analyse the side effect of chemotherapy in cancer patients. I used a chi2 test, it worked fine for most chemotherapy protocoles except for some chemotherapy regimen when I have a small number of patients (between 1 and 4). My question is: what is the best and suitable test to use and how do I analyse the outcome?

  Thank you for your help

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Marta Garcia-Granero]

Raouf Alami escribió:
> Hello everybody,
>   I would like to analyse the side effect of chemotherapy in cancer patients. I used a chi2 test, it worked fine for most chemotherapy protocoles except for some chemotherapy regimen when I have a small number of patients (between 1 and 4). My question is: what is the best and suitable test to use and how do I analyse the outcome?
>
I DON'T think that you can do a lot with that sample size. Try exact
tests (Fisher...), but don't expect any significant result, those tests
will lack power to detect even very large effects.

Regards,
Marta

--
For miscellaneous statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Lim Yonghao]

It might not be necessary to conduct a statistical test to answer your question (anyway, as Marta pointed out, the power will be so low that nothing will turn up significant).

If you suspect that administration of the chemotherapy protocol might result in some side effects and you observed them in your patients, i think that is good enough information. To strengthen your inference, you might want to see whether the side effects go away when the chemotherapy is being withdrawn (which might have ethical issues).

Hopefully, that will warrant enough attention for you to seriously consider a larger investigation.

Cheers,
Yonghao


> Date: Mon, 15 Sep 2008 11:05:12 +0200> From: [hidden email]> Subject: Re: When N is less than 5......?> To: [hidden email]> > Raouf Alami escribió:> > Hello everybody,> > I would like to analyse the side effect of chemotherapy in cancer patients. I used a chi2 test, it worked fine for most chemotherapy protocoles except for some chemotherapy regimen when I have a small number of patients (between 1 and 4). My question is: what is the best and suitable test to use and how do I analyse the outcome?> >> I DON'T think that you can do a lot with that sample size. Try exact> tests (Fisher...), but don't expect any significant result, those tests> will lack power to detect even very large effects.> > Regards,> Marta> > --> For miscellaneous statistical stuff, visit:> http://gjyp.nl/marta/> > =====================> To manage your subscription tto SPSSX-L, send a message to> [hidden email] (not to SPSSX-L), with no body text except the> command. To leave the list, send the command> SIGNOFF SPSSX-L> For a list of commands to manage subscriptions, send the command> INFO REFCARD
_________________________________________________________________
Manage multiple email accounts with Windows Live Mail effortlessly.
http://www.get.live.com/wl/all

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD