Posted by
Mike on
Jul 14, 2016; 7:10pm
URL: http://spssx-discussion.165.s1.nabble.com/futility-analysis-tp5732731p5732735.html
To add to what David says below:
(1) Futility analysis is not a statistical procedure (like doing a
t-test)
but a research strategy (e.g., half way through a study one does a
preliminary analysis to determine if the empirical result or, more
specifically, empirical effect size, is equal to a an expected
value or larger than the expected value [a superiority analysis];
if the empirical effect size is smaller than the expected or the
effect is equal to zero (supports the null hypothesis), this is a
futility analysis, that is, it futile to continue the Randomized
Control Trial (RCT; usually done in medical studies). The
futility analysis is a reason to stop the RCT and keep the
Type I error rate down (presumably one has sufficient
statistical power).
(2) To the OP: a futility/superiority/interim analysis would have
been built into the RCT design which identified what design
was being used and what analysis/analyses would be done.
The OP or someone associated with the RCT should either
have the grant application or the RCT plan that would provide
this information. The only appropriate question for the OP here
is whether certain statistical analyses can be conducted,
namely, those specified in in analysis plan. If the OP does
not know of such a plan, the OP needs to explain what exactly
he/she is doing.
(3) For a debate on the utility of futility analysis, see the
following:
http://ccforum.biomedcentral.com/articles/10.1186/cc3013A search of scholar.google.com and/or books.google.com will
turn up numerous sources for the rationale, philosophy, and
procedures (e.g., one-tailed test favored over two-tailed) that
are typically used.
The answers the OP wants are elsewhere.
-Mike Palij
New York University
[hidden email]
----- Original Message -----
From: "David Marso" <
[hidden email]>
To: <
[hidden email]>
Sent: Thursday, July 14, 2016 1:56 PM
Subject: Re: futility analysis
> Without further information from OP the quest IS futile.
> I searched online and it appears to be some Clinical Trials related
> interim
> analysis thing with many potential rabbit holes to tumble into.
> Not my area and I have truly no interest in becoming an expert in such
> matters so I will refrain from playing Alice today.
>
>
http://www.ncbi.nlm.nih.gov/pubmed/18283075> and
>
http://gozips.uakron.edu/~rps/Biostat%20Presentations%20Sp05/Interim%20Analysis-Apr15.pdf>
> Depending upon design and a load of other things, proceeding with
> further
> feedback is unwise aside from doing some research into Power analysis
> and
> ascertain what the appropriate analytics apply to analyzing the data
> in the
> first place. Looks complex and you should probably consult with a
> skilled
> statistician because you are not going to get any sort of one size
> fits all
> hand holding exercise from a newsgroup.
> OR break your own problem down into a set of manageable questions and
> ask
> them piecewise.
>
>
> Maguin, Eugene wrote
>> There has to be a clever pun or twist of meaning to play on the
>> futility
>> of asking people who may not or do not know what the term means how
>> to do
>> a futility analysis. Please tell us what a futility analysis is lest
>> your
>> quest be futile.
>> Gene Maguin
>>
>> From: SPSSX(r) Discussion [mailto:
>
>> SPSSX-L@.UGA
>
>> ] On Behalf Of Eun Um
>> Sent: Thursday, July 14, 2016 12:24 PM
>> To:
>
>> SPSSX-L@.UGA
>
>> Subject: futility analysis
>>
>> Can you let me know how I can conduct futility analysis in SPSS step
>> by
>> step? Thanks.
=====================
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