| View previous topic :: View next topic |
| Author |
Message |
Bicho the Inhaler
Daedalian Member
|
Posted: Tue Apr 08, 2003 9:25 pm Post subject: 1 |
|
|
I can't remember my statistics training! Suppose I take two samples, consisting say of N1 and N2 individuals, and I find that M1 individuals of the N1 sample have property Q, and M2 individuals of the N2 sample have property Q. How do I determine if the difference between the two samples is significant? Is this the t-test? I don't remember exactly how to do this type of test; when the results are valid; etc.
Thanks in advance. |
|
| Back to top |
|
 |
Ghost Post
Icarian Member
|
Posted: Tue Apr 08, 2003 10:41 pm Post subject: 2 |
|
|
no, it's a two sample z test
p=proportion with Q
z= p1-p2
sqrt(ppooled(1-ppooled)(1/n1+1/n2)) |
|
| Back to top |
|
 |
Ghost Post
Icarian Member
|
Posted: Tue Apr 08, 2003 10:58 pm Post subject: 3 |
|
|
| you look up the Z value you get on a table to see the probability of result that or more extreme. You need to decide in advance how small the probability should be to be significant, usually 5% or 1%. |
|
| Back to top |
|
 |
Bicho the Inhaler
Daedalian Member
|
Posted: Wed Apr 09, 2003 12:03 am Post subject: 4 |
|
|
| Ha ha, that looks vaguely familiar. Thanks, RSA. |
|
| Back to top |
|
 |
Ghost Post
Icarian Member
|
Posted: Wed Apr 09, 2003 12:11 am Post subject: 5 |
|
|
| no problem, the stats ap is fast approaching so this stuff is at the forefront of the ol' mind |
|
| Back to top |
|
 |
Legion
Daedalian Member
|
Posted: Thu Apr 10, 2003 10:58 am Post subject: 6 |
|
|
| Yeah, also Excel does t-test, p-test, a-test for you! But you gotta know haow to drive the thing. |
|
| Back to top |
|
 |
|