Allstat members,
Elementary Statistics - I HOPE
As a chemist I would appreciate any suggestions how to prove or disprove the
following as statistically significant or not significant.
Treatment on 122 "experimental" patients gives result A 28 times.
Another treatment on 122 "control" patients gives result A 20 times.
Is this SIGNIFICANT?
What methods of analysis are open to me?
The same experiments also gave at the same time another lesser result B (not
quite enough to be classed as result A) of 15 and 10 respectively.
Can this make the significance stronger?
Overall the results could be expressed as
122 gave results A + B of 43
and
122 gave results of A + B of 35
A "success" would be to get the experimental numbers below those of the
control.
The choice of 122 experimental patients may have been "biased" to do this.
Patients were certainly not chosen to give the high numbers that were found
in any event.
It was claimed the results do not show the first method to be any worse than
the normal treatment..
What established methods are available to test this?
What new techniques are available?
Are there any design faults to prevent any useful or meaningful statistics
to be done with these figures?
Thank you
John Fryer Chemist with "statistical enthusiasm"!
This is too important not to have some "professional" input or at least some
other views.
---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.393 / Virus Database: 223 - Release Date: 30/09/2002
|